We're halfway there with Extended Attributes and FSevents. We just need a decent framework for preserving metadata and and of course a common reference for managing tag behavior.
Those are manageable, they still have the potential to fail in way too many of the corner cases though. And those unresolved corner cases are the kind of thing Apple will refuse to ship something over, because if it doesn't work for everything they will get flayed and roasted alive across the whole internet.
Those are manageable, they still have the potential to fail in way too many of the corner cases though. And those unresolved corner cases are the kind of thing Apple will refuse to ship something over, because if it doesn't work for everything they will get flayed and roasted alive across the whole internet.
Yup which is why I don't expect Apple to move on this. I'm supporting OpenMeta because it's in a way the "proof of concept" but right now it's suitable for people who aren't afraid to risk the problems in order to glean the benefits.
I think both hierarchical folder structure and Tags will play well together. In fact I plan to leverage tagging to reduce the amount of folders total in my hierarchy but I don't expect it to eradicate the need for folder structure.
So let me get this straight. Apple gets panned for "adding" functionality in other apps and thus "killing" 3rd party players and now you want Apple to "kill" a 3rd party player because you don't want to pay beyond the $129 and get that add-on functionality?
Not at all. I paid for Leap because it's that good. I don't know why you're insinuating that I wouldn't want to pay beyond $129 to get such functionality. May I ask why?
Apple wouldn't kill Leap because unlike hierarchical file browser/managers, there are hundreds of different ways to implement a browser that makes use of metadata tags. Leap has it's particular way of handling them...if Apple wants to do it Leap's way, it can buy Ironic out...if it doesn't, it can bring some new way of visualizing tags.
edit: btw, you still haven't told me how OpenCL 1.0 depends on OpenGL 3.1.
This display of ignorance is illuminating. You want it all, want it right, want changed things to seem the same and have some things that are the same seem different. You want stuff to only show up once even though it may appear on the disk many times in identical formats because the filesystem should know they are the same. You want things that are different because we only changed little things to seem the same because semantically they are. You need this whole new paradigm but are restricting us to using Save and Save As.
What you want is not a bad thing, I want that too. But you absolutely refuse to believe the level of difficulty of the task when it has been laid out repeatedly. Get this through your thick skull, the more obvious subtle things are for people, the exponentially harder they are to program in a way that is executable on a reasonable time-scale. It is not frakking simple even though the idea is simple, because the idea is infused with intent and common sense, the hardest frakking nut there is to crack in computing.
Dude, like I said, I don't care how Apple does it. If it's really that complicated to implement at the filesystem level then Apple can simply have the OS hide the filesystem stuff and give the user a richer file browsing/managing experience with current filesystems like it's doing now with Spotlight.
It's not fuckin' voodoo...files with the same name that aren't identical can coexist on the same storage device as long as they're not in the same directory. Whooptee doo. If the filesystem can't easily handle two files with the same name in the same directory, it's not very difficult to imagine that the OS could create a directory for this file behind the scenes but still display it in, say, the Finder as a file in the same directory as the other file with the same name. I think users and the OS would be intelligent enough to know that they're not the same file considering the user would have have access to other metadata that would make it obvious that it's not the same file and the OS would know because it created a special directory for it so that the dumb filesystem doesn't spaz out.
I'm not asking for a huge paradigm shift that would move us away from Save and Save As...I'm just asking for some way for the OS to automate file managing by tagging files appropriately and giving the user a break from file micromanagement.
It's being done with music files, video files, photo files...now it just needs to be extended to *all* files and moved into the Finder.
If people are so fucking attached to the desktop metaphor and spatial management then why is a file name getting so much importance? Sit down at your desk, look at the papers and documents on your desk...do you find the one you want by reading the name of the document? I'm pretty such most documents don't even have a name. You look at the content of the document. The dimensions of the document. The color of the document.
If people are going to argue that spatial organization via hierarchical filesystems is still the way to go, then they should at least concede that finding a file by its file name is dumb as shit.
I fully embrace Coverflow and Spotlight. Some people love to pooh-pooh these two technologies but it's THE way to find documents. It's like asking a secretary to find the files for you no matter where they are. That's how things should be. It shouldn't matter where the files are or what the name is. You shouldn't have to file these things yourself. The secretary fetches them without error.
Anyway...I don't give a shit. Leap works well. The Finder can be made to work well with some trickery but it still involves a lot of micromanaging.
I'm not asking for a huge paradigm shift that would move us away from Save and Save As...I'm just asking for some way for the OS to automate file managing by tagging files appropriately and giving the user a break from file micromanagement.
That's implied by the system though. If you have a file open in say Photoshop and want to replace a file that already existed, how do you find that file in a dialog without a hierarchy that points uniquely to that file?
Quote:
Originally Posted by kim kap sol
If people are so fucking attached to the desktop metaphor and spatial management then why is a file name getting so much importance? Sit down at your desk, look at the papers and documents on your desk...do you find the one you want by reading the name of the document? I'm pretty such most documents don't even have a name. You look at the content of the document. The dimensions of the document. The color of the document.
Look at a filing cabinet and you can see that when it comes to large numbers of documents, it's just infeasible to look at content to find what you need. You flip through labels instead - same with a rolodex or address book. These systems do have limitations such as a rolodex not allowing you to find all people you know in a certain geographical radius but you simply don't need to know that - labeling fits its purpose and when it doesn't, a search can be tacked onto a digital equivalent.
Quote:
Originally Posted by kim kap sol
If people are going to argue that spatial organization via hierarchical filesystems is still the way to go, then they should at least concede that finding a file by its file name is dumb as shit.
Then I guess that every school is like that too by addressing students by name in assembly or every worker by their name when it comes to handing out paychecks. When you meet someone for the first time, the very first thing you exchange is typically name, you don't say 'hi there, I'm 6ft tall and I like to watch TV a lot, what's your shoe size?' You exchange one label that summarizes a person's relative uniqueness - name and extension (first name and surname).
Sure you note down a lot more information in your head like facial features, body size, hair color etc but when you address those people in order to uniquely identify them from a crowd, you don't shout 'hey big knockers, can you come in here?', you use their first name. Otherwise you just get fat Phil from accounting in your office when you want the secretary.
Quote:
Originally Posted by kim kap sol
I fully embrace Coverflow and Spotlight. Some people love to pooh-pooh these two technologies but it's THE way to find documents.
You missed something here. It's THE way that you find documents. That doesn't make it a universally better system for people who don't.
The fact that you can fit all your documents in one folder alone means you don't have enough to find the flaws in that system. When you try managing 10 times more files, the system shows it's flaws just like itunes does loading a large library.
Some people look at the Piles demoes and think it's a great way to manage files and it probably is for about 50 files but try it with 20,000 files and the setup is worthless.
That's implied by the system though. If you have a file open in say Photoshop and want to replace a file that already existed, how do you find that file in a dialog without a hierarchy that points uniquely to that file?
Because the OS or the smarter filesystem would know. I don't know why I keep explaining this. A file path is *not* the only way to uniquely identify a file.
Quote:
Look at a filing cabinet and you can see that when it comes to large numbers of documents, it's just infeasible to look at content to find what you need. You flip through labels instead - same with a rolodex or address book. These systems do have limitations such as a rolodex not allowing you to find all people you know in a certain geographical radius but you simply don't need to know that - labeling fits its purpose and when it doesn't, a search can be tacked onto a digital equivalent.
So you want computers to mimic real world rigid sorting/categorizing mechanisms when they can offer easier ways to sort and categorize? Marvin, I really don't know why you're using a computer. Perhaps you should be writing these posts on paper and sending them through the mail to everyone because you clearly couldn't care less that computers can offer more efficient ways to perform certain actions than what the physical world would allow.
Quote:
Then I guess that every school is like that too by addressing students by name in assembly or every worker by their name when it comes to handing out paychecks. When you meet someone for the first time, the very first thing you exchange is typically name, you don't say 'hi there, I'm 6ft tall and I like to watch TV a lot, what's your shoe size?' You exchange one label that summarizes a person's relative uniqueness - name and extension (first name and surname).
You got this wrong. While it's true that we're a society that pins a name or an identification number on people to enforce uniqueness (a flaw of our society), I think most would agree that when you meet someone for the first time and you exchange names, people will usually walk away and forget the name but remember the physical features. I'd love to see a study about this...I'm convinced most people simply remember that a person was tall, or short, or fat, or skinny, or spoke funny, or had a nice soothing voice, or was always smiling, or grumpy...but you ask them what the person's name was and most people would have forgotten it after a couple weeks.
Quote:
Sure you note down a lot more information in your head like facial features, body size, hair color etc but when you address those people in order to uniquely identify them from a crowd, you don't shout 'hey big knockers, can you come in here?', you use their first name. Otherwise you just get fat Phil from accounting in your office when you want the secretary.
Sure but you'd first have to remember someone's name. And if there are two Phils with the same last name, you have to go by other metadata such as weight. It's much easier to say 'hey big knockers, can you come in here?' and then pick from 3 people, 'fat Phil', 'Big Bertha', 'Dolly Parton'. The you can pick the secretary out of these 3.
Quote:
You missed something here. It's THE way that you find documents. That doesn't make it a universally better system for people who don't.
Right...and that's why I keep repeating that virtual hierarchies can still be created. So it's a win-win situation.
Quote:
The fact that you can fit all your documents in one folder alone means you don't have enough to find the flaws in that system. When you try managing 10 times more files, the system shows it's flaws just like itunes does loading a large library.
But those aren't flaws that are intrinsically tied to the concept, they're flaws tied to the OS, the app and the hardware.
Quote:
Some people look at the Piles demoes and think it's a great way to manage files and it probably is for about 50 files but try it with 20,000 files and the setup is worthless.
Piles, on the other hand, have flaws tied to the concept...just like file hierarchies. Because no amounts of OS performance tweaking, app tweaking, or faster hardware will allow users to find files within a hierarchies or within a pile any faster as the amount of files grow.
edit: Hierarchies were definitely the way to go back in the 70s considering the hardware in the 70s and 80s would not have been able to handle a more flexible structure. But we're past this. We have fast hardware. Signatures created from the metadata can be stamped onto files so that the filesystem knows it's a unique file or not. Filesystems don't need to be so dumb anymore.
Because the OS or the smarter filesystem would know. I don't know why I keep explaining this. A file path is *not* the only way to uniquely identify a file.
The point I'm making is that you need to tell the machine a lot of this required information in order that it knows which file you're talking about.
The scenario is that you have a file that has no predefined path with a name tree.jpg. In Photoshop, you create a new file with the intention of overwriting tree.jpg. When you click save as, how do you tell the OS that you want to overwrite tree.jpg if you can have files with the same names everywhere? All it would do is give it unique creation date metadata and co-exist with the file you wanted to replace.
Not only that even if you could replace files, every dialog would have to give you an image preview or something to that effect to make sure you were absolutely sure you were overwriting the file you wanted. A rigid hierarchy ensures there is no uncertainty because only one file with that name in that location can ever be that file. It's a 100% guarantee that non-linear identifiers don't give you.
Quote:
Originally Posted by kim kap sol
you clearly couldn't care less that computers can offer more efficient ways to perform certain actions than what the physical world would allow.
If there was a more efficient AND practical way then I would certainly care. I would agree that the meta-system in certain circumstances would be more efficient, mostly when it comes to media sorting. The issue is practicality. I have never seen a practical implementation of a meta-data system beyond what we already use, which is a rigid location structure with meta-data for advanced searches.
Quote:
Originally Posted by kim kap sol
I think most would agree that when you meet someone for the first time and you exchange names, people will usually walk away and forget the name but remember the physical features.
I would say that's likely but what I was saying is it's not a system that e.g. allows you identify them reliably to someone else. If you meet someone for the first time in your life for just 5 minutes and they introduced themselves by name and someone came up to you and asked who you were talking to, what would you say?
You would start with the name and if that didn't register, you would start to describe the physical characteristics that you thought were the most prominent. In a filesystem, this is equivalent to looking for a file directly by name and location as it's a simple, common unique identifier and failing that test, you then search using whatever metadata you are aware of.
Quote:
Originally Posted by kim kap sol
Right...and that's why I keep repeating that virtual hierarchies can still be created. So it's a win-win situation.
I know but the more you bring back the hierarchies out of necessity, the less need there is to replace them outright. There's no sense breaking a system that has worked for decades without many complaints for a system that has to impose many of the same conventions in order to be usable.
Quote:
Originally Posted by kim kap sol
But those aren't flaws that are intrinsically tied to the concept, they're flaws tied to the OS, the app and the hardware.
Correct but these limitations place constraints on when, not if this will happen. You were saying that such a thing should be in 10.6. Has hardware reached a level where this is feasible? Assuming 10.6 is Intel-only then it would mean that they have a minimum hardware spec to try and cope with, which is a big advantage moving forward but I still think we need a new filesystem and ubiquitous SSD drives for a system like that to be anything more than an extra feature.
I would reckon 10.7 is a good bet but I wouldn't expect it for a while after 10.6. I think that ZFS in 10.6 server could reveal more about what to expect later on though.
Quote:
Originally Posted by kim kap sol
Signatures created from the metadata can be stamped onto files so that the filesystem knows it's a unique file or not. Filesystems don't need to be so dumb anymore.
That's ok for the OS identifying a file itself but users won't use or remember file signatures so given that the OS knows all the metadata anyway, an extra signature probably wouldn't improve the issues.
The point I'm making is that you need to tell the machine a lot of this required information in order that it knows which file you're talking about.
The scenario is that you have a file that has no predefined path with a name tree.jpg. In Photoshop, you create a new file with the intention of overwriting tree.jpg. When you click save as, how do you tell the OS that you want to overwrite tree.jpg if you can have files with the same names everywhere? All it would do is give it unique creation date metadata and co-exist with the file you wanted to replace.
Umm, you would find the file you want to replace? Just like you would with today's filesystem? If it's crucial to overwrite the file, instead of navigating to the directory containing said file, you'd run a query using a pleasant interface such as Leap's interface and then select the file you want to overwrite. Ooooo...sooo difficult.
Quote:
Not only that even if you could replace files, every dialog would have to give you an image preview or something to that effect to make sure you were absolutely sure you were overwriting the file you wanted. A rigid hierarchy ensures there is no uncertainty because only one file with that name in that location can ever be that file. It's a 100% guarantee that non-linear identifiers don't give you.
Right, everyone would *automatically* know that they're overwriting the right file by looking at the path and the file name...riiiiight.
That said, we're not stuck in the 70s and 80s anymore where visual previews of files was near impossible (especially on CLI systems). So I don't see why the world couldn't just do away with ridiculous list views when pictures tell a thousand words.
Remember, list views are relics of the 70s. It wasn't uncommon to have hundreds of files in a directory and the OS and hardware simply couldn't give the user much more than the list of file names, extensions, and a few other metadata elements.
Today, it's possible to have visual previews of just a dozen files by filtering a search.
Quote:
If there was a more efficient AND practical way then I would certainly care. I would agree that the meta-system in certain circumstances would be more efficient, mostly when it comes to media sorting. The issue is practicality. I have never seen a practical implementation of a meta-data system beyond what we already use, which is a rigid location structure with meta-data for advanced searches.
Of course, you've never seen a practical implementation of a meta-data system beyond what you already use...hardly anyone is trying. Only a few small and daring independent developers are experimenting with the idea. I just wish Apple had the balls to do it. Apple isn't the same it was back in the 80s...the Apple of today hardly takes any risks.
Quote:
I would say that's likely but what I was saying is it's not a system that e.g. allows you identify them reliably to someone else. If you meet someone for the first time in your life for just 5 minutes and they introduced themselves by name and someone came up to you and asked who you were talking to, what would you say?
You would start with the name and if that didn't register, you would start to describe the physical characteristics that you thought were the most prominent. In a filesystem, this is equivalent to looking for a file directly by name and location as it's a simple, common unique identifier and failing that test, you then search using whatever metadata you are aware of.
Yes, I fully agree that I would start with the name and fall back on physical characteristics failing to remember the name. However, I think in most cases I would fall back on physical characteristics as time passed and as I slowly forget something as insignificant as a name.
You may have a superhuman memory but most people don't. You can't expect someone to remember thousands of names a few months later after meeting people for 5 minutes just like you can't expect people to remember file names or file paths a few months later after working on them for 5 minutes. You can definitely expect them to more easily remember them by other elements.
Quote:
I know but the more you bring back the hierarchies out of necessity, the less need there is to replace them outright. There's no sense breaking a system that has worked for decades without many complaints for a system that has to impose many of the same conventions in order to be usable.
Correct but these limitations place constraints on when, not if this will happen. You were saying that such a thing should be in 10.6. Has hardware reached a level where this is feasible? Assuming 10.6 is Intel-only then it would mean that they have a minimum hardware spec to try and cope with, which is a big advantage moving forward but I still think we need a new filesystem and ubiquitous SSD drives for a system like that to be anything more than an extra feature.
I would reckon 10.7 is a good bet but I wouldn't expect it for a while after 10.6. I think that ZFS in 10.6 server could reveal more about what to expect later on though.
Yes, I agree that it might be slightly early for such a system to work...but I also think you're right that we're getting close to having the hardware and the software to support such a system.
Quote:
That's ok for the OS identifying a file itself but users won't use or remember file signatures so given that the OS knows all the metadata anyway, an extra signature probably wouldn't improve the issues.
True...the signature would be invisible to the user and only used by the OS to warn the user of potential collision between two identical files or to keep tabs on what files had seen modifications so that the user knows that he can Save to overwrite or Save As to create a new file.
instead of navigating to the directory containing said file, you'd run a query using a pleasant interface such as Leap's interface and then select the file you want to overwrite. Ooooo...sooo difficult.
It's not the difficulty in finding a file to match a query but the certainty you have that the query returned the file you wanted without opening it or having to see a preview of it.
Quote:
Originally Posted by kim kap sol
Right, everyone would *automatically* know that they're overwriting the right file by looking at the path and the file name...riiiiight.
Generally they would but there's also the icon preview and the dates in the current setup. The question is how much would a revolutionary system really offer for day-to-day work?
Quote:
Originally Posted by kim kap sol
Apple of today hardly takes any risks.
Like introducing a mobile device in a market they haven't competed in with an interface completely different from every other device in that market? I would say Apple don't take unnecessary risks.
Quote:
Originally Posted by kim kap sol
However, I think in most cases I would fall back on physical characteristics as time passed and as I slowly forget something as insignificant as a name.
You may have a superhuman memory but most people don't. You can't expect someone to remember thousands of names a few months later after meeting people for 5 minutes just like you can't expect people to remember file names or file paths a few months later after working on them for 5 minutes. You can definitely expect them to more easily remember them by other elements.
There's a simple test to see: write down say 20 or more names of your family and friends on a piece of paper in a column. Then make a few other columns with attributes like height, eye color, hair color, age, shoe size or whatever attributes you would remember. Now try filling those columns in and check the amount you manage to get bearing in mind how long it takes to do so and also the fact that you've known these people for years.
Now try and do the same for your work colleagues and acquaintances.
Now with the information on that paper, imagine that you are looking up details of those people on your computer where it has records of them all. Which information are you going to use to search for them?
Marvin gets how important capturing a user's intention is to a significant advance in filesystem design.
What you and Marvin fail to do is explain why capturing user intentions is so important right now. I don't think the next logical step for filesystem is to read people's minds, Hiro. Are you really suggesting that? Stop fuckin' around.
It's not the difficulty in finding a file to match a query but the certainty you have that the query returned the file you wanted without opening it or having to see a preview of it.
Again, we're not living in the 70s or 80s anymore. Last time I checked it was possible to have large previews of pictures, documents, and listen to audio files without opening a file first.
Quote:
Generally they would but there's also the icon preview and the dates in the current setup. The question is how much would a revolutionary system really offer for day-to-day work?
So now you acknowledge that file previews exist to sway the argument to your side. Anyway, I think it offers a lot. I think Leopard is halfway there. I've never been more productive using a crippled Spotlight and Coverflow to get me to the right document...all this usually within 5-10 seconds. I don't have to think about the file name or the file path of a file I haven't touched for months...I just remember the general sense of the document, type one to three keywords (none now when I use Leap) and *bang*, I usually land on the document right away or have narrowed the search to a few documents if I typed only one search element.
I guarantee it would take *you* longer to find a file if it was one or two folders deep.
Quote:
Like introducing a mobile device in a market they haven't competed in with an interface completely different from every other device in that market? I would say Apple don't take unnecessary risks.
The risk was introducing an MP3 player 8 years ago...since then, they haven't taken significant risks. It's been smooth sailing since then.
Quote:
There's a simple test to see: write down say 20 or more names of your family and friends on a piece of paper in a column. Then make a few other columns with attributes like height, eye color, hair color, age, shoe size or whatever attributes you would remember. Now try filling those columns in and check the amount you manage to get bearing in mind how long it takes to do so and also the fact that you've known these people for years.
Now try and do the same for your work colleagues and acquaintances.
Now with the information on that paper, imagine that you are looking up details of those people on your computer where it has records of them all. Which information are you going to use to search for them?
I'm not sure what you're trying to say here. Are you saying that files you know are easier to look up by name and files you don't know very well are easier to look up by attribute or content?
I'm just not seeing the roadblocks that you all see with tagging.
The benefit of tagging with something like OpenMeta is that you and only you
create OpenMeta tags so really it comes down to how people view documents and
their constituent "attributes"
I may simply tag groups by "friends" or "family" and thus I've created a tagging distinction between David my cousin and David my friend.
Leap is really good at removing folder structure and frankly it's a bit of a shock to view your application folder and see ALL your files. Folders organize and maintain to a certain point but they also obsfuscate your view of your documents.
Tagging attempts to give you a holistic view and then you drill down based on metadata not folder structure. I think this is certainly a better and more flexible arrangement for people that need to track disparate files for projects or case files.
What you and Marvin fail to do is explain why capturing user intentions is so important right now. I don't think the next logical step for filesystem is to read people's minds, Hiro. Are you really suggesting that? Stop fuckin' around.
Conveying your intent to a machine is essential for it to do any job correctly as it only has the most basic inputs. Mind reading wasn't implied, the problem is simply creating a system that guarantees you are telling the machine exactly what you want. As mentioned, a fixed hierarchy guarantees that no file can possibly exist in that location besides the file with a given name. A complete meta-system as you described removes that certainty.
Quote:
Originally Posted by kim kap sol
Last time I checked it was possible to have large previews of pictures, documents, and listen to audio files without opening a file first.
Usually but what happens when you have a file type the filebrowser can't generate previews for such as .iff files or perhaps .mkv files. You are using previews to make up for losing one measure of certainty gained from a rigid folder structure.
Quote:
Originally Posted by kim kap sol
So now you acknowledge that file previews exist to sway the argument to your side.
I'm saying that the current system has the previews that you are suggesting to rely on as well as a folder structure. The more information the better IMO but it should be conveyed as concisely as possible. My concern is over the conciseness that your suggested system would present the same information already on offer in the current system - in some ways, it seems like a downgraded system vs the hierarchy + meta-data.
Quote:
Originally Posted by kim kap sol
I'm not sure what you're trying to say here. Are you saying that files you know are easier to look up by name and files you don't know very well are easier to look up by attribute or content?
Yes and you seem to agree as your example was:
"I don't have to think about the file name or the file path of a file I haven't touched for months"
This is the most effective use for meta-data. Your guarantees about finding things in known locations faster using meta-data suggest you disagree but day-to-day, I personally find my files faster using a hierarchy. I can see this being mostly a performance constraint but there's also something to be said about the ease with which files are tagged as well as the effectiveness of the tags.
If I tag a file as belonging to project x but I also used it in project y, I could easily forget I used it in project y. I then proceed to work on this file having found it via a query to find images related to project x. I modify the file. Later on, I do a query to find files by project y only to find that I've now changed a file that I need unmodified. A hierarchy would have prevented this mistake because project x and y files are maintained in separate hierarchies so I would have made sure to create duplicate files.
Having a hierarchy + meta-data still allows me to group them in a query but it prevents mistakes.
Quote:
Originally Posted by hmurchison
I'm just not seeing the roadblocks that you all see with tagging.
The benefit of tagging with something like OpenMeta is that you and only you
create OpenMeta tags so really it comes down to how people view documents and
their constituent "attributes"
The question isn't whether or not tagging is beneficial or not. It certainly is. The question is does it need to replace a rigid hierarchical folder structure instead of being used alongside it?
The benefits you mention of a user being left to tag items can also be a big flaw because on a multi-user system not everyone will expect the same tags, not everyone will bother to add tags and it can lead to a messy filesystem. This can happen with fixed structures but it almost enforces the use of tagging to maintain order.
The benefits you mention of a user being left to tag items can also be a big flaw because on a multi-user system not everyone will expect the same tags, not everyone will bother to add tags and it can lead to a messy filesystem. This can happen with fixed structures but it almost enforces the use of tagging to maintain order.
Oh yes certainly, tagging can be personal or it can follow established rules but for shared volumes folder hierarchy is going to be a more sane solution with a bit of tagging over the top to help things out. I think we need both.
Tagging to reduce folder proliferation
Folder to ease the burden of "too many tags" syndrome.
All I want to say though (after deliberately erasing my entire post when I realized I'm just restating things over and over again) is that a file name is just another piece of metadata. It shouldn't have more importance or less importance than other metadata. The most important thing about a file is the data itself -- the content. The metadata should always be secondary.
I understand that some people have grown to think the file name is the most important thing in the world and that it would be difficult to find a file without a file name. That, however, is bullshit. It's an artificial constraint that people have accepted as "the way it has to be".
I guess I can squeeze in one last comment: Hiro is a bastard for changing this into a filesystem conversation. It was never meant to be one. He took a single thing I said and blew it up to epic proportions. Damn
Let me restate it...I don't care what kind of filesystem sits behind the OS. I just want Apple to work on something similar to Leap. Apple can keep HFS+ in the background for all I care as long as they start giving more importance to its Spotlight tech.
Too bad the conversation you started wasn't the rant you wanted to leave it as. And the ENTIRE topic was about the filesystem! The display of a filesystems contents cannot ever be separated from the metadata or how he metadata is managed, that's a by-definition kind of thing. Well unless you like displaying random content, then we can safely separate the two sides of the issue.
Comments
We're halfway there with Extended Attributes and FSevents. We just need a decent framework for preserving metadata and and of course a common reference for managing tag behavior.
Those are manageable, they still have the potential to fail in way too many of the corner cases though. And those unresolved corner cases are the kind of thing Apple will refuse to ship something over, because if it doesn't work for everything they will get flayed and roasted alive across the whole internet.
Yup which is why I don't expect Apple to move on this. I'm supporting OpenMeta because it's in a way the "proof of concept" but right now it's suitable for people who aren't afraid to risk the problems in order to glean the benefits.
I think both hierarchical folder structure and Tags will play well together. In fact I plan to leverage tagging to reduce the amount of folders total in my hierarchy but I don't expect it to eradicate the need for folder structure.
So let me get this straight. Apple gets panned for "adding" functionality in other apps and thus "killing" 3rd party players and now you want Apple to "kill" a 3rd party player because you don't want to pay beyond the $129 and get that add-on functionality?
Not at all. I paid for Leap because it's that good. I don't know why you're insinuating that I wouldn't want to pay beyond $129 to get such functionality. May I ask why?
Apple wouldn't kill Leap because unlike hierarchical file browser/managers, there are hundreds of different ways to implement a browser that makes use of metadata tags. Leap has it's particular way of handling them...if Apple wants to do it Leap's way, it can buy Ironic out...if it doesn't, it can bring some new way of visualizing tags.
edit: btw, you still haven't told me how OpenCL 1.0 depends on OpenGL 3.1.
This display of ignorance is illuminating. You want it all, want it right, want changed things to seem the same and have some things that are the same seem different. You want stuff to only show up once even though it may appear on the disk many times in identical formats because the filesystem should know they are the same. You want things that are different because we only changed little things to seem the same because semantically they are. You need this whole new paradigm but are restricting us to using Save and Save As.
What you want is not a bad thing, I want that too. But you absolutely refuse to believe the level of difficulty of the task when it has been laid out repeatedly. Get this through your thick skull, the more obvious subtle things are for people, the exponentially harder they are to program in a way that is executable on a reasonable time-scale. It is not frakking simple even though the idea is simple, because the idea is infused with intent and common sense, the hardest frakking nut there is to crack in computing.
Dude, like I said, I don't care how Apple does it. If it's really that complicated to implement at the filesystem level then Apple can simply have the OS hide the filesystem stuff and give the user a richer file browsing/managing experience with current filesystems like it's doing now with Spotlight.
It's not fuckin' voodoo...files with the same name that aren't identical can coexist on the same storage device as long as they're not in the same directory. Whooptee doo. If the filesystem can't easily handle two files with the same name in the same directory, it's not very difficult to imagine that the OS could create a directory for this file behind the scenes but still display it in, say, the Finder as a file in the same directory as the other file with the same name. I think users and the OS would be intelligent enough to know that they're not the same file considering the user would have have access to other metadata that would make it obvious that it's not the same file and the OS would know because it created a special directory for it so that the dumb filesystem doesn't spaz out.
I'm not asking for a huge paradigm shift that would move us away from Save and Save As...I'm just asking for some way for the OS to automate file managing by tagging files appropriately and giving the user a break from file micromanagement.
It's being done with music files, video files, photo files...now it just needs to be extended to *all* files and moved into the Finder.
If people are so fucking attached to the desktop metaphor and spatial management then why is a file name getting so much importance? Sit down at your desk, look at the papers and documents on your desk...do you find the one you want by reading the name of the document? I'm pretty such most documents don't even have a name. You look at the content of the document. The dimensions of the document. The color of the document.
If people are going to argue that spatial organization via hierarchical filesystems is still the way to go, then they should at least concede that finding a file by its file name is dumb as shit.
I fully embrace Coverflow and Spotlight. Some people love to pooh-pooh these two technologies but it's THE way to find documents. It's like asking a secretary to find the files for you no matter where they are. That's how things should be. It shouldn't matter where the files are or what the name is. You shouldn't have to file these things yourself. The secretary fetches them without error.
Anyway...I don't give a shit. Leap works well. The Finder can be made to work well with some trickery but it still involves a lot of micromanaging.
I'm not asking for a huge paradigm shift that would move us away from Save and Save As...I'm just asking for some way for the OS to automate file managing by tagging files appropriately and giving the user a break from file micromanagement.
That's implied by the system though. If you have a file open in say Photoshop and want to replace a file that already existed, how do you find that file in a dialog without a hierarchy that points uniquely to that file?
If people are so fucking attached to the desktop metaphor and spatial management then why is a file name getting so much importance? Sit down at your desk, look at the papers and documents on your desk...do you find the one you want by reading the name of the document? I'm pretty such most documents don't even have a name. You look at the content of the document. The dimensions of the document. The color of the document.
Look at a filing cabinet and you can see that when it comes to large numbers of documents, it's just infeasible to look at content to find what you need. You flip through labels instead - same with a rolodex or address book. These systems do have limitations such as a rolodex not allowing you to find all people you know in a certain geographical radius but you simply don't need to know that - labeling fits its purpose and when it doesn't, a search can be tacked onto a digital equivalent.
If people are going to argue that spatial organization via hierarchical filesystems is still the way to go, then they should at least concede that finding a file by its file name is dumb as shit.
Then I guess that every school is like that too by addressing students by name in assembly or every worker by their name when it comes to handing out paychecks. When you meet someone for the first time, the very first thing you exchange is typically name, you don't say 'hi there, I'm 6ft tall and I like to watch TV a lot, what's your shoe size?' You exchange one label that summarizes a person's relative uniqueness - name and extension (first name and surname).
Sure you note down a lot more information in your head like facial features, body size, hair color etc but when you address those people in order to uniquely identify them from a crowd, you don't shout 'hey big knockers, can you come in here?', you use their first name. Otherwise you just get fat Phil from accounting in your office when you want the secretary.
I fully embrace Coverflow and Spotlight. Some people love to pooh-pooh these two technologies but it's THE way to find documents.
You missed something here. It's THE way that you find documents. That doesn't make it a universally better system for people who don't.
The fact that you can fit all your documents in one folder alone means you don't have enough to find the flaws in that system. When you try managing 10 times more files, the system shows it's flaws just like itunes does loading a large library.
Some people look at the Piles demoes and think it's a great way to manage files and it probably is for about 50 files but try it with 20,000 files and the setup is worthless.
That's implied by the system though. If you have a file open in say Photoshop and want to replace a file that already existed, how do you find that file in a dialog without a hierarchy that points uniquely to that file?
Because the OS or the smarter filesystem would know. I don't know why I keep explaining this. A file path is *not* the only way to uniquely identify a file.
Look at a filing cabinet and you can see that when it comes to large numbers of documents, it's just infeasible to look at content to find what you need. You flip through labels instead - same with a rolodex or address book. These systems do have limitations such as a rolodex not allowing you to find all people you know in a certain geographical radius but you simply don't need to know that - labeling fits its purpose and when it doesn't, a search can be tacked onto a digital equivalent.
So you want computers to mimic real world rigid sorting/categorizing mechanisms when they can offer easier ways to sort and categorize? Marvin, I really don't know why you're using a computer. Perhaps you should be writing these posts on paper and sending them through the mail to everyone because you clearly couldn't care less that computers can offer more efficient ways to perform certain actions than what the physical world would allow.
Then I guess that every school is like that too by addressing students by name in assembly or every worker by their name when it comes to handing out paychecks. When you meet someone for the first time, the very first thing you exchange is typically name, you don't say 'hi there, I'm 6ft tall and I like to watch TV a lot, what's your shoe size?' You exchange one label that summarizes a person's relative uniqueness - name and extension (first name and surname).
You got this wrong. While it's true that we're a society that pins a name or an identification number on people to enforce uniqueness (a flaw of our society), I think most would agree that when you meet someone for the first time and you exchange names, people will usually walk away and forget the name but remember the physical features. I'd love to see a study about this...I'm convinced most people simply remember that a person was tall, or short, or fat, or skinny, or spoke funny, or had a nice soothing voice, or was always smiling, or grumpy...but you ask them what the person's name was and most people would have forgotten it after a couple weeks.
Sure you note down a lot more information in your head like facial features, body size, hair color etc but when you address those people in order to uniquely identify them from a crowd, you don't shout 'hey big knockers, can you come in here?', you use their first name. Otherwise you just get fat Phil from accounting in your office when you want the secretary.
Sure but you'd first have to remember someone's name. And if there are two Phils with the same last name, you have to go by other metadata such as weight. It's much easier to say 'hey big knockers, can you come in here?' and then pick from 3 people, 'fat Phil', 'Big Bertha', 'Dolly Parton'. The you can pick the secretary out of these 3.
You missed something here. It's THE way that you find documents. That doesn't make it a universally better system for people who don't.
Right...and that's why I keep repeating that virtual hierarchies can still be created. So it's a win-win situation.
The fact that you can fit all your documents in one folder alone means you don't have enough to find the flaws in that system. When you try managing 10 times more files, the system shows it's flaws just like itunes does loading a large library.
But those aren't flaws that are intrinsically tied to the concept, they're flaws tied to the OS, the app and the hardware.
Some people look at the Piles demoes and think it's a great way to manage files and it probably is for about 50 files but try it with 20,000 files and the setup is worthless.
Piles, on the other hand, have flaws tied to the concept...just like file hierarchies. Because no amounts of OS performance tweaking, app tweaking, or faster hardware will allow users to find files within a hierarchies or within a pile any faster as the amount of files grow.
edit: Hierarchies were definitely the way to go back in the 70s considering the hardware in the 70s and 80s would not have been able to handle a more flexible structure. But we're past this. We have fast hardware. Signatures created from the metadata can be stamped onto files so that the filesystem knows it's a unique file or not. Filesystems don't need to be so dumb anymore.
Because the OS or the smarter filesystem would know. I don't know why I keep explaining this. A file path is *not* the only way to uniquely identify a file.
The point I'm making is that you need to tell the machine a lot of this required information in order that it knows which file you're talking about.
The scenario is that you have a file that has no predefined path with a name tree.jpg. In Photoshop, you create a new file with the intention of overwriting tree.jpg. When you click save as, how do you tell the OS that you want to overwrite tree.jpg if you can have files with the same names everywhere? All it would do is give it unique creation date metadata and co-exist with the file you wanted to replace.
Not only that even if you could replace files, every dialog would have to give you an image preview or something to that effect to make sure you were absolutely sure you were overwriting the file you wanted. A rigid hierarchy ensures there is no uncertainty because only one file with that name in that location can ever be that file. It's a 100% guarantee that non-linear identifiers don't give you.
you clearly couldn't care less that computers can offer more efficient ways to perform certain actions than what the physical world would allow.
If there was a more efficient AND practical way then I would certainly care. I would agree that the meta-system in certain circumstances would be more efficient, mostly when it comes to media sorting. The issue is practicality. I have never seen a practical implementation of a meta-data system beyond what we already use, which is a rigid location structure with meta-data for advanced searches.
I think most would agree that when you meet someone for the first time and you exchange names, people will usually walk away and forget the name but remember the physical features.
I would say that's likely but what I was saying is it's not a system that e.g. allows you identify them reliably to someone else. If you meet someone for the first time in your life for just 5 minutes and they introduced themselves by name and someone came up to you and asked who you were talking to, what would you say?
You would start with the name and if that didn't register, you would start to describe the physical characteristics that you thought were the most prominent. In a filesystem, this is equivalent to looking for a file directly by name and location as it's a simple, common unique identifier and failing that test, you then search using whatever metadata you are aware of.
Right...and that's why I keep repeating that virtual hierarchies can still be created. So it's a win-win situation.
I know but the more you bring back the hierarchies out of necessity, the less need there is to replace them outright. There's no sense breaking a system that has worked for decades without many complaints for a system that has to impose many of the same conventions in order to be usable.
But those aren't flaws that are intrinsically tied to the concept, they're flaws tied to the OS, the app and the hardware.
Correct but these limitations place constraints on when, not if this will happen. You were saying that such a thing should be in 10.6. Has hardware reached a level where this is feasible? Assuming 10.6 is Intel-only then it would mean that they have a minimum hardware spec to try and cope with, which is a big advantage moving forward but I still think we need a new filesystem and ubiquitous SSD drives for a system like that to be anything more than an extra feature.
I would reckon 10.7 is a good bet but I wouldn't expect it for a while after 10.6. I think that ZFS in 10.6 server could reveal more about what to expect later on though.
Signatures created from the metadata can be stamped onto files so that the filesystem knows it's a unique file or not. Filesystems don't need to be so dumb anymore.
That's ok for the OS identifying a file itself but users won't use or remember file signatures so given that the OS knows all the metadata anyway, an extra signature probably wouldn't improve the issues.
The point I'm making is that you need to tell the machine a lot of this required information in order that it knows which file you're talking about.
The scenario is that you have a file that has no predefined path with a name tree.jpg. In Photoshop, you create a new file with the intention of overwriting tree.jpg. When you click save as, how do you tell the OS that you want to overwrite tree.jpg if you can have files with the same names everywhere? All it would do is give it unique creation date metadata and co-exist with the file you wanted to replace.
Umm, you would find the file you want to replace? Just like you would with today's filesystem? If it's crucial to overwrite the file, instead of navigating to the directory containing said file, you'd run a query using a pleasant interface such as Leap's interface and then select the file you want to overwrite. Ooooo...sooo difficult.
Not only that even if you could replace files, every dialog would have to give you an image preview or something to that effect to make sure you were absolutely sure you were overwriting the file you wanted. A rigid hierarchy ensures there is no uncertainty because only one file with that name in that location can ever be that file. It's a 100% guarantee that non-linear identifiers don't give you.
Right, everyone would *automatically* know that they're overwriting the right file by looking at the path and the file name...riiiiight.
That said, we're not stuck in the 70s and 80s anymore where visual previews of files was near impossible (especially on CLI systems). So I don't see why the world couldn't just do away with ridiculous list views when pictures tell a thousand words.
Remember, list views are relics of the 70s. It wasn't uncommon to have hundreds of files in a directory and the OS and hardware simply couldn't give the user much more than the list of file names, extensions, and a few other metadata elements.
Today, it's possible to have visual previews of just a dozen files by filtering a search.
If there was a more efficient AND practical way then I would certainly care. I would agree that the meta-system in certain circumstances would be more efficient, mostly when it comes to media sorting. The issue is practicality. I have never seen a practical implementation of a meta-data system beyond what we already use, which is a rigid location structure with meta-data for advanced searches.
Of course, you've never seen a practical implementation of a meta-data system beyond what you already use...hardly anyone is trying. Only a few small and daring independent developers are experimenting with the idea. I just wish Apple had the balls to do it. Apple isn't the same it was back in the 80s...the Apple of today hardly takes any risks.
I would say that's likely but what I was saying is it's not a system that e.g. allows you identify them reliably to someone else. If you meet someone for the first time in your life for just 5 minutes and they introduced themselves by name and someone came up to you and asked who you were talking to, what would you say?
You would start with the name and if that didn't register, you would start to describe the physical characteristics that you thought were the most prominent. In a filesystem, this is equivalent to looking for a file directly by name and location as it's a simple, common unique identifier and failing that test, you then search using whatever metadata you are aware of.
Yes, I fully agree that I would start with the name and fall back on physical characteristics failing to remember the name. However, I think in most cases I would fall back on physical characteristics as time passed and as I slowly forget something as insignificant as a name.
You may have a superhuman memory but most people don't. You can't expect someone to remember thousands of names a few months later after meeting people for 5 minutes just like you can't expect people to remember file names or file paths a few months later after working on them for 5 minutes. You can definitely expect them to more easily remember them by other elements.
I know but the more you bring back the hierarchies out of necessity, the less need there is to replace them outright. There's no sense breaking a system that has worked for decades without many complaints for a system that has to impose many of the same conventions in order to be usable.
Correct but these limitations place constraints on when, not if this will happen. You were saying that such a thing should be in 10.6. Has hardware reached a level where this is feasible? Assuming 10.6 is Intel-only then it would mean that they have a minimum hardware spec to try and cope with, which is a big advantage moving forward but I still think we need a new filesystem and ubiquitous SSD drives for a system like that to be anything more than an extra feature.
I would reckon 10.7 is a good bet but I wouldn't expect it for a while after 10.6. I think that ZFS in 10.6 server could reveal more about what to expect later on though.
Yes, I agree that it might be slightly early for such a system to work...but I also think you're right that we're getting close to having the hardware and the software to support such a system.
That's ok for the OS identifying a file itself but users won't use or remember file signatures so given that the OS knows all the metadata anyway, an extra signature probably wouldn't improve the issues.
True...the signature would be invisible to the user and only used by the OS to warn the user of potential collision between two identical files or to keep tabs on what files had seen modifications so that the user knows that he can Save to overwrite or Save As to create a new file.
instead of navigating to the directory containing said file, you'd run a query using a pleasant interface such as Leap's interface and then select the file you want to overwrite. Ooooo...sooo difficult.
It's not the difficulty in finding a file to match a query but the certainty you have that the query returned the file you wanted without opening it or having to see a preview of it.
Right, everyone would *automatically* know that they're overwriting the right file by looking at the path and the file name...riiiiight.
Generally they would but there's also the icon preview and the dates in the current setup. The question is how much would a revolutionary system really offer for day-to-day work?
Apple of today hardly takes any risks.
Like introducing a mobile device in a market they haven't competed in with an interface completely different from every other device in that market? I would say Apple don't take unnecessary risks.
However, I think in most cases I would fall back on physical characteristics as time passed and as I slowly forget something as insignificant as a name.
You may have a superhuman memory but most people don't. You can't expect someone to remember thousands of names a few months later after meeting people for 5 minutes just like you can't expect people to remember file names or file paths a few months later after working on them for 5 minutes. You can definitely expect them to more easily remember them by other elements.
There's a simple test to see: write down say 20 or more names of your family and friends on a piece of paper in a column. Then make a few other columns with attributes like height, eye color, hair color, age, shoe size or whatever attributes you would remember. Now try filling those columns in and check the amount you manage to get bearing in mind how long it takes to do so and also the fact that you've known these people for years.
Now try and do the same for your work colleagues and acquaintances.
Now with the information on that paper, imagine that you are looking up details of those people on your computer where it has records of them all. Which information are you going to use to search for them?
Marvin gets how important capturing a user's intention is to a significant advance in filesystem design.
What you and Marvin fail to do is explain why capturing user intentions is so important right now. I don't think the next logical step for filesystem is to read people's minds, Hiro. Are you really suggesting that? Stop fuckin' around.
It's not the difficulty in finding a file to match a query but the certainty you have that the query returned the file you wanted without opening it or having to see a preview of it.
Again, we're not living in the 70s or 80s anymore. Last time I checked it was possible to have large previews of pictures, documents, and listen to audio files without opening a file first.
Generally they would but there's also the icon preview and the dates in the current setup. The question is how much would a revolutionary system really offer for day-to-day work?
So now you acknowledge that file previews exist to sway the argument to your side. Anyway, I think it offers a lot. I think Leopard is halfway there. I've never been more productive using a crippled Spotlight and Coverflow to get me to the right document...all this usually within 5-10 seconds. I don't have to think about the file name or the file path of a file I haven't touched for months...I just remember the general sense of the document, type one to three keywords (none now when I use Leap) and *bang*, I usually land on the document right away or have narrowed the search to a few documents if I typed only one search element.
I guarantee it would take *you* longer to find a file if it was one or two folders deep.
Like introducing a mobile device in a market they haven't competed in with an interface completely different from every other device in that market? I would say Apple don't take unnecessary risks.
The risk was introducing an MP3 player 8 years ago...since then, they haven't taken significant risks. It's been smooth sailing since then.
There's a simple test to see: write down say 20 or more names of your family and friends on a piece of paper in a column. Then make a few other columns with attributes like height, eye color, hair color, age, shoe size or whatever attributes you would remember. Now try filling those columns in and check the amount you manage to get bearing in mind how long it takes to do so and also the fact that you've known these people for years.
Now try and do the same for your work colleagues and acquaintances.
Now with the information on that paper, imagine that you are looking up details of those people on your computer where it has records of them all. Which information are you going to use to search for them?
I'm not sure what you're trying to say here. Are you saying that files you know are easier to look up by name and files you don't know very well are easier to look up by attribute or content?
The benefit of tagging with something like OpenMeta is that you and only you
create OpenMeta tags so really it comes down to how people view documents and
their constituent "attributes"
I may simply tag groups by "friends" or "family" and thus I've created a tagging distinction between David my cousin and David my friend.
Leap is really good at removing folder structure and frankly it's a bit of a shock to view your application folder and see ALL your files. Folders organize and maintain to a certain point but they also obsfuscate your view of your documents.
Tagging attempts to give you a holistic view and then you drill down based on metadata not folder structure. I think this is certainly a better and more flexible arrangement for people that need to track disparate files for projects or case files.
What you and Marvin fail to do is explain why capturing user intentions is so important right now. I don't think the next logical step for filesystem is to read people's minds, Hiro. Are you really suggesting that? Stop fuckin' around.
Conveying your intent to a machine is essential for it to do any job correctly as it only has the most basic inputs. Mind reading wasn't implied, the problem is simply creating a system that guarantees you are telling the machine exactly what you want. As mentioned, a fixed hierarchy guarantees that no file can possibly exist in that location besides the file with a given name. A complete meta-system as you described removes that certainty.
Last time I checked it was possible to have large previews of pictures, documents, and listen to audio files without opening a file first.
Usually but what happens when you have a file type the filebrowser can't generate previews for such as .iff files or perhaps .mkv files. You are using previews to make up for losing one measure of certainty gained from a rigid folder structure.
So now you acknowledge that file previews exist to sway the argument to your side.
I'm saying that the current system has the previews that you are suggesting to rely on as well as a folder structure. The more information the better IMO but it should be conveyed as concisely as possible. My concern is over the conciseness that your suggested system would present the same information already on offer in the current system - in some ways, it seems like a downgraded system vs the hierarchy + meta-data.
I'm not sure what you're trying to say here. Are you saying that files you know are easier to look up by name and files you don't know very well are easier to look up by attribute or content?
Yes and you seem to agree as your example was:
"I don't have to think about the file name or the file path of a file I haven't touched for months"
This is the most effective use for meta-data. Your guarantees about finding things in known locations faster using meta-data suggest you disagree but day-to-day, I personally find my files faster using a hierarchy. I can see this being mostly a performance constraint but there's also something to be said about the ease with which files are tagged as well as the effectiveness of the tags.
If I tag a file as belonging to project x but I also used it in project y, I could easily forget I used it in project y. I then proceed to work on this file having found it via a query to find images related to project x. I modify the file. Later on, I do a query to find files by project y only to find that I've now changed a file that I need unmodified. A hierarchy would have prevented this mistake because project x and y files are maintained in separate hierarchies so I would have made sure to create duplicate files.
Having a hierarchy + meta-data still allows me to group them in a query but it prevents mistakes.
I'm just not seeing the roadblocks that you all see with tagging.
The benefit of tagging with something like OpenMeta is that you and only you
create OpenMeta tags so really it comes down to how people view documents and
their constituent "attributes"
The question isn't whether or not tagging is beneficial or not. It certainly is. The question is does it need to replace a rigid hierarchical folder structure instead of being used alongside it?
The benefits you mention of a user being left to tag items can also be a big flaw because on a multi-user system not everyone will expect the same tags, not everyone will bother to add tags and it can lead to a messy filesystem. This can happen with fixed structures but it almost enforces the use of tagging to maintain order.
The benefits you mention of a user being left to tag items can also be a big flaw because on a multi-user system not everyone will expect the same tags, not everyone will bother to add tags and it can lead to a messy filesystem. This can happen with fixed structures but it almost enforces the use of tagging to maintain order.
Oh yes certainly, tagging can be personal or it can follow established rules but for shared volumes folder hierarchy is going to be a more sane solution with a bit of tagging over the top to help things out. I think we need both.
Tagging to reduce folder proliferation
Folder to ease the burden of "too many tags" syndrome.
All I want to say though (after deliberately erasing my entire post when I realized I'm just restating things over and over again) is that a file name is just another piece of metadata. It shouldn't have more importance or less importance than other metadata. The most important thing about a file is the data itself -- the content. The metadata should always be secondary.
I understand that some people have grown to think the file name is the most important thing in the world and that it would be difficult to find a file without a file name. That, however, is bullshit. It's an artificial constraint that people have accepted as "the way it has to be".
Anyway...byez, Marvin and Hiro.
Let me restate it...I don't care what kind of filesystem sits behind the OS. I just want Apple to work on something similar to Leap. Apple can keep HFS+ in the background for all I care as long as they start giving more importance to its Spotlight tech.
Too bad the conversation you started wasn't the rant you wanted to leave it as. And the ENTIRE topic was about the filesystem! The display of a filesystems contents cannot ever be separated from the metadata or how he metadata is managed, that's a by-definition kind of thing. Well unless you like displaying random content, then we can safely separate the two sides of the issue.