I think in the scheme things it will all work out. A global pandemic kind of puts these things in to perspective
Did people not see this issue coming? I mean at some point a 2 digit agent identify isn’t going to work any more. I know Safari allows you to change agent identification, surely Firefox could emulate an older versions automatically if needed.
I don't see this as a browser issue. I see this as a lack of forward thinking on the part of web servers that only accept 2 digit ids. Let the web server developers get their acts together.
When Microsoft releases a new OS upgrade, you can be sure they test hundreds of apps on hundreds of different PCs (tens of thousands of combinations) before they release the OS upgrade. Even though Microsoft doesn't have the legal responsibility to ensure compatibility, I really appreciate that they try. Browser developers should do the same thing, because browsers are akin to an OS for so many people and purposes.
This about WEB servers not an OS problem. Web servers run an app on a PC that is internet facing. It 's the app that needs updating.
You didn't get my comparison. Because I agree with you that what you say is legally true, but for moral reasons Microsoft does its best to ensure compatibility, and for the same moral reasons browser developers should try to ensure compatibility. In fact, some browsers have had commands in their menus that let you, the user, report compatibility issues to them. Are you saying browsers should stop doing that and stop worrying about compatibility?
I think in the scheme things it will all work out. A global pandemic kind of puts these things in to perspective
Did people not see this issue coming? I mean at some point a 2 digit agent identify isn’t going to work any more. I know Safari allows you to change agent identification, surely Firefox could emulate an older versions automatically if needed.
I don't see this as a browser issue. I see this as a lack of forward thinking on the part of web servers that only accept 2 digit ids. Let the web server developers get their acts together.
When Microsoft releases a new OS upgrade, you can be sure they test hundreds of apps on hundreds of different PCs (tens of thousands of combinations) before they release the OS upgrade. Even though Microsoft doesn't have the legal responsibility to ensure compatibility, I really appreciate that they try. Browser developers should do the same thing, because browsers are akin to an OS for so many people and purposes.
This about WEB servers not an OS problem. Web servers run an app on a PC that is internet facing. It 's the app that needs updating.
You didn't get my comparison. Because I agree with you that what you say is legally true, but for moral reasons Microsoft does its best to ensure compatibility, and for the same moral reasons browser developers should try to ensure compatibility. In fact, some browsers have had commands in their menus that let you, the user, report compatibility issues to them. Are you saying browsers should stop doing that and stop worrying about compatibility?
I think in the scheme things it will all work out. A global pandemic kind of puts these things in to perspective
Did people not see this issue coming? I mean at some point a 2 digit agent identify isn’t going to work any more. I know Safari allows you to change agent identification, surely Firefox could emulate an older versions automatically if needed.
I don't see this as a browser issue. I see this as a lack of forward thinking on the part of web servers that only accept 2 digit ids. Let the web server developers get their acts together.
When Microsoft releases a new OS upgrade, you can be sure they test hundreds of apps on hundreds of different PCs (tens of thousands of combinations) before they release the OS upgrade. Even though Microsoft doesn't have the legal responsibility to ensure compatibility, I really appreciate that they try. Browser developers should do the same thing, because browsers are akin to an OS for so many people and purposes.
This about WEB servers not an OS problem. Web servers run an app on a PC that is internet facing. It 's the app that needs updating.
You didn't get my comparison. Because I agree with you that what you say is legally true, but for moral reasons Microsoft does its best to ensure compatibility, and for the same moral reasons browser developers should try to ensure compatibility. In fact, some browsers have had commands in their menus that let you, the user, report compatibility issues to them. Are you saying browsers should stop doing that and stop worrying about compatibility?
Moral reasons?
Not everything is about legality. Platform developers like OS devs and Browser devs might simply want users to have a good experience.
Besides if a website breaks on one browser and not another, even if it's that site's fault, who do you think the user will blame?
I think in the scheme things it will all work out. A global pandemic kind of puts these things in to perspective
Did people not see this issue coming? I mean at some point a 2 digit agent identify isn’t going to work any more. I know Safari allows you to change agent identification, surely Firefox could emulate an older versions automatically if needed.
I don't see this as a browser issue. I see this as a lack of forward thinking on the part of web servers that only accept 2 digit ids. Let the web server developers get their acts together.
When Microsoft releases a new OS upgrade, you can be sure they test hundreds of apps on hundreds of different PCs (tens of thousands of combinations) before they release the OS upgrade. Even though Microsoft doesn't have the legal responsibility to ensure compatibility, I really appreciate that they try. Browser developers should do the same thing, because browsers are akin to an OS for so many people and purposes.
This about WEB servers not an OS problem. Web servers run an app on a PC that is internet facing. It 's the app that needs updating.
You didn't get my comparison. Because I agree with you that what you say is legally true, but for moral reasons Microsoft does its best to ensure compatibility, and for the same moral reasons browser developers should try to ensure compatibility. In fact, some browsers have had commands in their menus that let you, the user, report compatibility issues to them. Are you saying browsers should stop doing that and stop worrying about compatibility?
Moral reasons?
Not everything is about legality. Platform developers like OS devs and Browser devs might simply want users to have a good experience.
Besides if a website breaks on one browser and not another, even if it's that site's fault, who do you think the user will blame?
Yeah no. Microsoft maintain compatibility because they own the APIs and have responsibility for them, and more importantly because compatibility is a selling point. Nothing to do with morality.
Browser developers don't own the technologies behind the internet, and have no knowledge or control over the crazy things that websites may have done. They can smoke test against some websites to ensure their browsers work, but if a web dev has hackily implemented something as simple as browser version control then the best thing would be for browsers to try and detect that (where practically possible) and to alert the user that the website is out of date. They could offer a temporary solution that allows the user to enter a compatibility mode that misreports the browser version number as 99 I guess, but that has to be a temporary solution. All this is user experience, and a selling point for your product, again nothing to do with morality.
*sigh* Y2K all over again. We found and fixed SO MANY things leading up to that. Mostly silly shortcuts that programmers took. They cut corners and it came back to bite them. Same thing here. Why did they give a fixed digit space for version. The article even said they ran into this going from single to two digit version numbers. They should have taken care of it once and for all.
No, nobody "took shortcuts" leading up to the Y2K thing:
Through the 70's, 80's and most of the 90's storage, particularly hardrive storage (called "DASD"), was severely limited and VERY expensive. An expensive, washing machine sized 3350 disk drive held 300Mb -- less than many mobile devices and PCs today (less than 5% that the iPhone 14 might hold). Think about storing payroll data for a large corporation with 30,000 employees on one of those things! And, to make things worse, each track of that device held 19,000 bytes -- so even the length of a record had to be carefully calculated and grouped into blocks: for a simple example: if you had a record 5,000 bytes long, you would waste 4,000 bytes of each track -- an unforgivable programming error that would get you chastised if not fired for incompetence.
So, every effort was made to insure that every one of of those 300Mb's was used wisely and not a single one was wasted. The result was that using a 2 digit year was a calculated decision to save storage and maximize computing power. Otherwise, applications would have to run from mag tape rather than harddrives which would bog down the entire process because the datasets would have to be accessed sequentially (meaning, for example, to look up employee number 9999 the computer would have to read through 9998 other employees to find him.)
By the year 2000 things had loosened up and DASD was both much cheaper and more plentiful. So, instead of changing the calculation, programmers added the other two digits onto the date so, when you subtracted 1998 from 2001 you didn't get (negative) -97.
This wouldnt be an issue if they didn't use such ridiculous versioning.
Do you know any versioning scheme that is less ridiculous that counting up: 1, 2, 3, .... ?
As far as I know, there is no standard for versioning, so I don’t know why any one company’s scheme would be any better than another. Within companies there is sometimes a standard model applied that implies compatibility or interoperability with file formats or data models, each of which have their own versioning scheme, but again, it doesn’t apply outside of the scope of the company.
Even in cases where there are standards in place it can often get a little (or a lot) ugly due to legacy formats that never seem to go away quickly enough. Two areas that always make me crazy are string formats and date-time formats. Date-time formats are particularly ugly because there are too many opinions about what actual date-time is the zero reference. On the string (and emoji) front, Unicode has made things a lot better, but there are still a lot of legacy applications and strings being emitted from embedded systems that don’t follow the Unicode standard.
There’s still a lot of ugliness lurking in code and every so often it leaks out to where regular people are exposed to it. Nature of the beast.
I think in the scheme things it will all work out. A global pandemic kind of puts these things in to perspective
Did people not see this issue coming? I mean at some point a 2 digit agent identify isn’t going to work any more. I know Safari allows you to change agent identification, surely Firefox could emulate an older versions automatically if needed.
I don't see this as a browser issue. I see this as a lack of forward thinking on the part of web servers that only accept 2 digit ids. Let the web server developers get their acts together.
Exactly. This is the opposite of the Y2K problem we had, except for the Mac. That was a really simple problem of the industry needing to save every bit they could in the first few decades, but apparently not thinking that dates would ever go past 1999. Here, we don’t have a digit problem in the application software, just in the passing of those digits in the websites. Another screwup. How could they not realize that numbering would go past 99.xxx? One would think that when it was reaching, say, 80, that the handwriting was on the wall, so to speak, and they would have begun upgrading it.
I remember that I had a big argument with a guy who was a developer and had a site where he was supposedly an “expert” on upgrade numbering standards. I told him that Apple would likely go to 10.10 and higher, and he became furious. According to the standards, he said, Apple couldn’t do that. Well, I told him the standards were bunk, and that Apple could do whatever they wanted to. He didn’t like that.
this isn’t a problem for something like that, but this site problem is different. Obviously all sides should have gotten together on this, but haven’t.
This is a much better versioning system. It summarises to the end user what has changed since the last update: bug fix, new feature, or major upgrade
Google and Mozilla’s versioning is stupid as it has exacerbated the 3 digit issue and is not informative to the end user.
I agree that Semantic Versioning is one of the better conventions being followed by many dev organizations. It's also very well spec'd out in BNF notation to make it more verifiable. Unfortunately, it's still not a standard as in something backed by an ISO/IEC/IEEE type of standards committee and many companies still do their own thing, as is evident in Google Chrome and a lot of Microsoft apps. The reality is that we may never get to a simple numerical standard like semantic versioning that covers all possible interoperability and dependency requirements. We may need some sort of manifest exchange based model with dynamic negotiation of compatibility, which would kind of suck for end users. I'd imagine that Apple's universal apps add another layer of complexity to the versioning challenge.
I'm not saying they deserve it, but maybe if they had used versioning numbers properly they wouldn't be in this mess
Or the web designers could have actually learned the lessons of YK2 and Windows 9x and futureproofed their web sites. Enough blame for this mess to go around.
*sigh* Y2K all over again. We found and fixed SO MANY things leading up to that. Mostly silly shortcuts that programmers took. They cut corners and it came back to bite them. Same thing here. Why did they give a fixed digit space for version. The article even said they ran into this going from single to two digit version numbers. They should have taken care of it once and for all.
No, nobody "took shortcuts" leading up to the Y2K thing:
Through the 70's, 80's and most of the 90's storage, particularly hardrive storage (called "DASD"), was severely limited and VERY expensive.
Epoch was a viable option and wouldn't cost much more than what they did do. For example the Classic MacOS epoch (beginning in 1984) started with January 1, 1904 and ended February 6, 2040 (though you could only set it to December 31, 2019) and Unix stared with January 1 1970. The 386 (32-bit) came out in 1985. Anybody still using only two digits by the times 90s rolled around was taking lazy shortcuts.
In fact, the less known 2020 bug was thanks to one of the fixes to the Y2K issue - windowing and was entirely a software problem not a hardware one.
I said decades ago that user agent strings being sent to websites was a stupid idea. Once again it's come back to bite people.
I blame MicroSloth for the current mess, of course. Had websites not sent different content to different browsers because of Internet Exploder and the MS attempt to embrace, extend, and extinguish, user agent strings would have been eliminated as the privacy problem they are decades ago.
*sigh* Y2K all over again. We found and fixed SO MANY things leading up to that. Mostly silly shortcuts that programmers took. They cut corners and it came back to bite them. Same thing here. Why did they give a fixed digit space for version. The article even said they ran into this going from single to two digit version numbers. They should have taken care of it once and for all.
No, nobody "took shortcuts" leading up to the Y2K thing:
Through the 70's, 80's and most of the 90's storage, particularly hardrive storage (called "DASD"), was severely limited and VERY expensive.
Epoch was a viable option and wouldn't cost much more than what they did do. For example the Classic MacOS epoch (beginning in 1984) started with January 1, 1904 and ended February 6, 2040 (though you could only set it to December 31, 2019) and Unix stared with January 1 1970. The 386 (32-bit) came out in 1985. Anybody still using only two digits by the times 90s rolled around was taking lazy shortcuts.
In fact, the less known 2020 bug was thanks to one of the fixes to the Y2K issue - windowing and was entirely a software problem not a hardware one.
Epoch is an equally bad option. Classic Mac OS has already hit part of the problem, UNIX will hit it in 2038.
At this point, the only sane fix is to stop that, and store an 11 digit year, then month, day, hour, minute, second, second fraction. Do it as an actual value, not a count of seconds from some arbitrary date. Yes, 11 digit year. Yes, that's because if you make it negative you can store any date back to the beginning of the universe, and any date that is likely to be relevant to humans in the future.
More storage by a little bit, problem solved permanently.
Comments
Besides if a website breaks on one browser and not another, even if it's that site's fault, who do you think the user will blame?
Browser developers don't own the technologies behind the internet, and have no knowledge or control over the crazy things that websites may have done. They can smoke test against some websites to ensure their browsers work, but if a web dev has hackily implemented something as simple as browser version control then the best thing would be for browsers to try and detect that (where practically possible) and to alert the user that the website is out of date. They could offer a temporary solution that allows the user to enter a compatibility mode that misreports the browser version number as 99 I guess, but that has to be a temporary solution. All this is user experience, and a selling point for your product, again nothing to do with morality.
Even in cases where there are standards in place it can often get a little (or a lot) ugly due to legacy formats that never seem to go away quickly enough. Two areas that always make me crazy are string formats and date-time formats. Date-time formats are particularly ugly because there are too many opinions about what actual date-time is the zero reference. On the string (and emoji) front, Unicode has made things a lot better, but there are still a lot of legacy applications and strings being emitted from embedded systems that don’t follow the Unicode standard.
There’s still a lot of ugliness lurking in code and every so often it leaks out to where regular people are exposed to it. Nature of the beast.
I remember that I had a big argument with a guy who was a developer and had a site where he was supposedly an “expert” on upgrade numbering standards. I told him that Apple would likely go to 10.10 and higher, and he became furious. According to the standards, he said, Apple couldn’t do that. Well, I told him the standards were bunk, and that Apple could do whatever they wanted to. He didn’t like that.
this isn’t a problem for something like that, but this site problem is different. Obviously all sides should have gotten together on this, but haven’t.
Google and Mozilla’s versioning is stupid as it has exacerbated the 3 digit issue and is not informative to the end user.
I agree that Semantic Versioning is one of the better conventions being followed by many dev organizations. It's also very well spec'd out in BNF notation to make it more verifiable. Unfortunately, it's still not a standard as in something backed by an ISO/IEC/IEEE type of standards committee and many companies still do their own thing, as is evident in Google Chrome and a lot of Microsoft apps. The reality is that we may never get to a simple numerical standard like semantic versioning that covers all possible interoperability and dependency requirements. We may need some sort of manifest exchange based model with dynamic negotiation of compatibility, which would kind of suck for end users. I'd imagine that Apple's universal apps add another layer of complexity to the versioning challenge.
In fact, the less known 2020 bug was thanks to one of the fixes to the Y2K issue - windowing and was entirely a software problem not a hardware one.