Bye, Windows 98

Discussion of general topics about Mozilla Firefox
Locked
User avatar
Thumper
Posts: 8037
Joined: November 4th, 2002, 5:42 pm
Location: Linlithgow, Scotland
Contact:

Bye, Windows 98

Post by Thumper »

The bug to stop maintaining Windows 98 compatibility on trunk was just closed. Woo.

Predictably, some idiot opened a counter-bug. if anyone in here has the appropriate privs, fancy wontfixing it?

(commentary on either of those bugs is a Federal offense, by the way)

- Chris
supernova_00
Posts: 4832
Joined: June 24th, 2004, 8:03 pm
Location: Maryland, USA

Post by supernova_00 »

About time! btw I marked that counter-bug as WONTFIX
The Ex Omega
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Post by The Ex Omega »

Its about damn time. Windows 98 is an old bird that needed retirement.
User avatar
BenBasson
Moderator
Posts: 13671
Joined: February 13th, 2004, 5:49 am
Location: London, UK
Contact:

Post by BenBasson »

Pre-emptive whine crush:

------- Comment #36 From Robert O'Callahan (Novell) 2006-03-29 21:07 PST [reply] -------

Here's what needs to be done to restore Win9x support, by whoever volunteers to
do so:

Define a new header file wincompat.h (or something like that). When building
normally it does nothing. When building for Win9x (enabled by a configure
flag), it #defines all the Win2K-only APIs to Win9x-based replacements. E.g.
PeekMessageW gets #defined to Win9x_PeekMessageW, which gets implemented in a
separate file as a wrapper around Win9x's PeekMessage. This patch is a great
place to start.

You should then be able to get Win9x-builds working without interfering with
trunk code. Regular Windows builds won't work on Win9x but that's no great
loss. I'd be happy to review such a patch.

Please don't:
- file new bugs asking for Win98 compatibility to be restored
- follow up on any developer blogs
- rant in the newsgroups or on this forum.

Please do:
- file a follow-up bug for the above behaviour and comment in 330276 stating that you have done so
- write a patch that implements it
- ask roc for review
- distribute third-party builds with Win98 support.
User avatar
BenBasson
Moderator
Posts: 13671
Joined: February 13th, 2004, 5:49 am
Location: London, UK
Contact:

Post by BenBasson »

A further addendum:

1) Mozilla developers are perfectly capable of considering the consequences of their actions. The bug wouldn't have been filed in the first place if they weren't intentionally dropping support for Windows 98 by Firefox 3.0. It wouldn't have been filed if there wasn't a general consensus that doing so is the right thing.

2) Firefox 3.0 is a year or more away. That means a year or more of official builds that work in Windows 98 and a year or more for people to organise a hacking community that produces working and up-to-date Windows 98 builds. The SeaMonkey project is managing pretty well. I suspect if enough people are interested and ask MoCo nicely, it might even be possible to borrow resources, but then again, finding a third-party builder shouldn't be amazingly difficult.

3) This is why open source is great. You can step up and maintain Firefox for your OS when your OS is eventually too old for mainstream developer time. With closed-source products, there's essentially no chance of this happening.
old zeniko
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Post by old zeniko »

supernova_00 wrote:About time! btw I marked that counter-bug as WONTFIX

OT: Although a pretty obvious WONTFIX, please don't do this unless you're module peer or owner.
User avatar
DizzyWeb
Posts: 637
Joined: March 27th, 2003, 9:56 am

Post by DizzyWeb »

I really don't understand what the problem is. Microsoft has dropped support for Win9x in their next version of IE. Do we hear complaints? No. By that time there will be Vista, so dropping support for something as ancient as Win9x is normal.

But if Mozilla announces to drop support for Firefox 3.0, which is 2 major releases away, a lot of people complain. Why?! Does Mozilla support Linux kernel 1.2 (from 1995) or version 2.0 (from 1998)? No. For obvious reasons.
The author can never, in no way, be held responsible for any harm caused, mental or physical, by reading this post.
User avatar
BenBasson
Moderator
Posts: 13671
Joined: February 13th, 2004, 5:49 am
Location: London, UK
Contact:

Post by BenBasson »

The counter-bug has been reopened by the bug reporter. I'm trying to bring the discussion here.
User avatar
BenBasson
Moderator
Posts: 13671
Joined: February 13th, 2004, 5:49 am
Location: London, UK
Contact:

Post by BenBasson »

Breaking Win98SE will disallow users to run a current Gecko on their old computers, and why should they run a browser only on their top machine, if they have one?

Nobody designs hardware or software with Windows 98 in-mind any more and they haven't done so for a while now. To whine about this is silly. It's outdated and difficult to work with in comparison to more recent releases. The option to not upgrade beyond Firefox 2.0 exists.

And the this bug is resolved as WONTFIX not by a module owner, or one of the
programmers involved in 'fixing' Bug 330276, no, Supernova is telling me this
bug is WONTFIX because the other one has been fixed. No Sir, as offered in
Bug 330276 Comment #36, RoC offers review if someone wants to make a shim
library supporting Win9x.
Bug 330276 Comment #35, RoC mentions concerns that that checkin also removed
some code supporting transparency on Win2k.

Unless the next thing posted to your bug is a patch, it's beyond ever getting taken seriously by a developer.

Maybe somebody feels fit to make a shim library like RoC suggested in Bug
330276 Comment #36

Your bug should be closed and a new bug should be opened *specifically* for that issue, describing it in a technically correct way, like "Implement wincompat.h header for Pre-Win2k compatibility" and quoting Roc and Darin's comments where appropriate. Any extra random "please support my OS" comments should be avoided. Bugzilla is tool, not a forum for discussion.
herman
Posts: 1034
Joined: November 7th, 2002, 3:45 pm

<b>I'm the idiot who opened the counter bug</b>

Post by herman »

<b>I'm the idiot who opened the counter bug</b>
and I feel pissed off by the silly non-technical argumentation.

The browsers on Win9x had been in a perfectly working condition, then somebody breaks it. What then follows seems to me like 'kindergarten', breaking a glass, then hiding the sherds, then telling: It was old, I don't like it, you don't need it anymore.

Imho you mustn't break win9x to run cairo. If you've got to break Win9x, then why not telling it in advance, telling technical reasons besides it's too old?

Putting the higly experimental, full of critical bugs, breaking at all ends software into the trunk, and hide the working software into the experimental branch seems to be logical? What are experimental branches for?
No, there are some checkins not intended to, but breaking Win9x, a week later a bug filed to support a warning on unsupported OS, yet a week later a bug to not support systems older than Win2k. Is that a 'planned decision'?

Imho Firefox is a browser for the masses of consumers, or why all that marketing fuzz? The masses never ran Win2k, they ran Win98SE until buying a computer running WinXP.
And now some kiddy is demanding dad should buy a shiny new computer running shiny new Vista to stay on top of browser technology, just for avoiding to be a stuoid old codger? A customer buying Vista will likely stick to IE7 until the memory leak issue of some Firefox Installations is sorted out.

Cusser wrote:
Breaking Win98SE will disallow users to run a current Gecko on their old computers, and why should they run a browser only on their top machine, if they have one?

Nobody designs hardware or software with Windows 98 in-mind any more and they haven't done so for a while now. To whine about this is silly. It's outdated and difficult to work with in comparison to more recent releases.

You are right that nobody designs Software with Win9x in mind.
But what are frameworks for? Does Firefox not run anymore on Win9x the moment MS stops supporting Win9x? Surely not. Is active work needed to support Win9x? Are there new problems to be solved?

Cusser wrote:The option to not upgrade beyond Firefox 2.0 exists.

That's adding insult to injury. Firefox2.0 will end like Mozilla 1.7.13, full of bugs, only security bugs fixed, with a recommendation to user newer software, if you want to avoid the bugs. No way, dead end street.
If I've got to accept bugs, I prefer those on trunk nightlies ;-)

I've got a lot of people switching from IE to some other browser, because I suggested the one fitting to them. I didn't try to convince everybody to use Mozilla, if they seemed a tad conservative, I succesfully suggested using Netscape7. Opera was better for some very bad-eyed people. There is no simple solution to all problems on a computer, and if it were, that surely won't be Firefox, but a new OS, maybe EMACS ;-). (ok, that's been the prove for the 'old codger')

I'm not strictly opposed to end support for Win9x, no. But it should be communicated better, in advance. I'm sure somebody knows how many Firefoxes are running on Win9x, or easily can get to know ;-)

It's a clandestine decision, like the decision not to release a Mozilla Suite 1.8, like the decision in <a href="https://bugzilla.mozilla.org/show_bug.cgi?id=203960">Bug 203960 Make bookmark groups conditionally replace existing tabs instead of appending</a>. In that one, the "conditional" part of the fix, i.e. the possibility to select the previous behaviour, was restored months later. The new behaviour not only lost the capability of loading multiple groupmarks, it also instantly produced about 10 bugs, still unfixed, leading to data-loss. The default preference is the buggy new behaviour, not the unbuggy old one. That bug was a crippling of the power of tabbed browsing, and the 'usability study' laying ground to that bug was never published.

Cusser wrote:
And then this bug is resolved as WONTFIX not by a module owner, or one of the programmers involved in 'fixing' Bug 330276, no, Supernova is telling me this bug is WONTFIX because the other one has been fixed.

Unless the next thing posted to your bug is a patch, it's beyond ever getting taken seriously by a developer.

Why should I post a next thing? I only wanted to tell the specialists that a WONTFIX decision is up to the module owner, and he knows when to WONTFIX it.
Cusser wrote:
Maybe somebody feels fit to make a shim library like RoC suggested in Bug
330276 Comment #36

Your bug should be closed and a new bug should be opened *specifically* for that issue, describing it in a technically correct way, like "Implement wincompat.h header for Pre-Win2k compatibility" and quoting Roc and Darin's comments where appropriate.

I'd added the "helpwanted" to tell that I'm not fit to do this myself. Anybody taking that bug would descibe it better than I could.
We don't live in a perfect world, bug-spamming is to be avoided, but always happens. Therefore look at this bug as a honey-pot bug, better people are commenting here than inappropriately whining in technical bugs. There are a lot of bugs open exactly for this reason.
Cusser wrote: Any extra random "please support my OS" comments should be avoided. Bugzilla is tool, not a forum for discussion.

I know, but a counter bug is a legit tool to avoid discussion in a technical bug, and I opened the counter bug to reduce discussion in the original technical bug to a less annoying level.
And there was a distinct difference in quality in the answers of developers, and the guys posting in this thread, if you like to reread both bugs.

<a href="https://bugzilla.mozilla.org/show_bug.cgi?id=330208"></a>2006-03-11 Bug 330208 Unsupported windows versions need error message on startup (and in installer)
<a href="https://bugzilla.mozilla.org/show_bug.cgi?id=330276">2006-03-12 Bug 330276 Drop support for pre-Win2k platforms</a>
<a href="https://bugzilla.mozilla.org/show_bug.cgi?id=331723">2006-03-25 Bug 331723 Don't kill Win98</a>
<a href=""></a>
<a href=""></a>
<a href=""></a>
herman
User avatar
BenBasson
Moderator
Posts: 13671
Joined: February 13th, 2004, 5:49 am
Location: London, UK
Contact:

Re: <b>I'm the idiot who opened the counter bug</b&

Post by BenBasson »

herman wrote:Imho you mustn't break win9x to run cairo. If you've got to break Win9x, then why not telling it in advance, telling technical reasons besides it's too old?

Sure thing - I asked Robert O'Callahan to comment via email.

Robert O'Callahan wrote:The main issue is that our new cairo graphics infrastructure does not work on Win9x, because it requires APIs only available in Win2K and later. It could be made to work on Win9x, probably, but that's extra work for developers who are already overloaded. Therefore Win9x support isn't going to happen unless someone else steps up to provide it.

Since trunk has switch to cairo, and trunk builds no longer work on Win9x, we've also taken the chance to remove a number of WIn9x compatibility hacks that were complicating our code. Those hacks could be put back in in a cleaner way that does not complicate our code, but again, this depends on someone who cares volunteering to do the work.


herman wrote:Putting the higly experimental, full of critical bugs, breaking at all ends software into the trunk, and hide the working software into the experimental branch seems to be logical? What are experimental branches for?
No, there are some checkins not intended to, but breaking Win9x, a week later a bug filed to support a warning on unsupported OS, yet a week later a bug to not support systems older than Win2k. Is that a 'planned decision'?

It's been known that Pre-Win2k support would be broken eventually for a while. There have been comments to that effect on the Mozilla wiki.

herman wrote:Imho Firefox is a browser for the masses of consumers, or why all that marketing fuzz? The masses never ran Win2k, they ran Win98SE until buying a computer running WinXP.

Yeah, they ran into Win98 eight years ago. Considering that you can no longer buy anything that works on Windows 98, I find it hard to believe that there's a consumer expectation that new free products continue to do so.

herman wrote:You are right that nobody designs Software with Win9x in mind.
But what are frameworks for? Does Firefox not run anymore on Win9x the moment MS stops supporting Win9x? Surely not. Is active work needed to support Win9x? Are there new problems to be solved?

Yes, active work is needed. Mozilla is in a constant state of improvement and there are a large number of complicated hacks and workarounds needed to support any improvements on Windows 98 because of it's age and API deficiencies. The only option is to either pay people to work on Windows 98 compatibility (which nobody is interested in doing, it seems) or leave it behind while moving forward on other platforms.

herman wrote:That's adding insult to injury. Firefox2.0 will end like Mozilla 1.7.13, full of bugs, only security bugs fixed, with a recommendation to user newer software, if you want to avoid the bugs. No way, dead end street.

Sorry, that's how it works. By the time Firefox 3.0 is released some time in 2007, I fully expect that a lot of people won't even have upgraded to Firefox 2.0 yet. The fact that you use nightlies on Windows 98 means that you're in an even smaller minority and also that you're out of luck.

herman wrote:I'm not strictly opposed to end support for Win9x, no. But it should be communicated better, in advance. I'm sure somebody knows how many Firefoxes are running on Win9x, or easily can get to know ;-)

Nobody is being secretive about it. You're using bleeding-edge builds. The official announcement will probably come some time before Firefox 3.0's release, not before the release of a random developer build.

herman wrote:It's a clandestine decision, like the decision not to release a Mozilla Suite 1.8

Actually, that was clearly stated at least a year in advance as part of the roadmap.

herman wrote:Why should I post a next thing? I only wanted to tell the specialists that a WONTFIX decision is up to the module owner, and he knows when to WONTFIX it.

It'll get WONTFIXED again because it's not addressing a specific problem that can be patched. The only useful thing you could do with it is turn it into a meta bug to track smaller, more specific bugs that when combined achieve compatibility.
djcater
Posts: 131
Joined: February 10th, 2005, 1:14 pm

Post by djcater »

herman, may I ask, where do you draw the line? How many missing APIs should Mozilla cater for? How hard does it have to become to support older OSs before they decide to drop support and clean up the code? Notice that older MacOS support was removed like this some time last year by Josh I believe. It has to happen for development to be practical.
User avatar
Amsterdammer
Posts: 752
Joined: July 7th, 2005, 1:10 pm
Location: Amsterdam, The Netherlands
Contact:

OS statistcs

Post by Amsterdammer »

look here

scroll to OS Platform Statistics
cannot find a better one at the moment

ys ;)
User avatar
BenBasson
Moderator
Posts: 13671
Joined: February 13th, 2004, 5:49 am
Location: London, UK
Contact:

Re: OS statistcs

Post by BenBasson »

Amsterdammer wrote:look here

scroll to OS Platform Statistics
cannot find a better one at the moment

ys ;)

Although market share is not that important to this discussion, W3Schools isn't a very good source. I found http://marketshare.hitslink.com/report.aspx?qprid=5 which likely more accurately describes market share.

After extrapolating the data using Excel, I found that at the current rate of decline:
- Windows 98 would have 0% market share by January 2007
- Windows ME would have 0% market share by April 2007
- Windows 95 would have 0% market share by June 2006 (!)
supernova_00
Posts: 4832
Joined: June 24th, 2004, 8:03 pm
Location: Maryland, USA

Re: <b>I'm the idiot who opened the counter bug</b&

Post by supernova_00 »

herman wrote:Why should I post a next thing? I only wanted to tell the specialists that a WONTFIX decision is up to the module owner, and he knows when to WONTFIX it.

It'll get WONTFIXED again because it's not addressing a specific problem that can be patched. The only useful thing you could do with it is turn it into a meta bug to track smaller, more specific bugs that when combined achieve compatibility.
I marked it WONTFIX to try and advoid the ever popular bug spam that plauges bugzilla when something like this happens. I could have (even though I'm not supposed to in the first place) just WONTFIXed it without referencing the guilty bug and stating partially why this happend. Also, Don't kill win-98 is a forum topic not a bug report. This should have started as a thread in the first place. Now like what cusser suggested is the way you should have went.

EDIT: and look at these stats from w3c

2006 WinXP W2000 Win98 WinNT W2003 Linux Mac
March 72.9% 11.9% 2.0% 0.3% 1.8% 3.4% 3.5%
February 73.3% 12.3% 2.1% 0.3% 1.8% 3.4% 3.6%

2005 WinXP W2000 Win98 WinNT W2003 Linux Mac
April 64.0% 19.7% 4.1% 0.8% 1.4% 3.3% 2.9%
March 63.1% 20.2% 4.7% 0.9% 1.4% 3.2% 3.0%

2004 WinXP W2000 Win98 WinNT Win95 Linux Mac
April 49.7% 30.2% 8.7% 2.2% 0.3% 2.7% 2.5%
March 48.0% 31.1% 9.4% 2.4% 0.4% 2.6% 2.4%

Win 98 has dropped 7 percent in last two years down to 2.0 percent last month. In theory that would mean 0 percent by 2008 or later when fx 3 is released. I'll give it 1% just for talk...so firefox should cater to 1% of the market? Uh no that is not how anything works in any type of business.
Locked