|
|
Subscribe / Log in / New account

Fedora, Mozilla, and trademarks

Fedora, Mozilla, and trademarks

Posted Apr 28, 2010 22:03 UTC (Wed) by roc (subscriber, #30627)
Parent article: Fedora, Mozilla, and trademarks

One thing we (Mozilla) should do is make it clear that any upstream patch approved for a branch landing (but perhaps not yet landed on that branch, or shipped in a release off that branch) is fair game for trademark-using distributors to apply to their own copy of that branch without further approval. This would help grease the wheels.

The MNG vs APNG issue is a joke. MNG is a huge bloaty spec that has never been implemented by any mainstream browser. It's grotesquely over-engineered --- it includes stuff like scaling of image chunks and linear gradients in its "image format". APNG is a much lighter extension to PNG. We submitted patches for APNG to libpng but the libpng owners rejected them because they like MNG. APNG is implemented in Opera as well as Firefox.

I think we would love people to submit patches to upstream to enable use of system libffi and other system libraries. As far as I know that hasn't happened yet.


to post comments

Fedora, Mozilla, and trademarks

Posted Apr 28, 2010 22:44 UTC (Wed) by jspaleta (subscriber, #50639) [Link] (2 responses)

Yes i think reading over the bulk of the thread... knowing when an upstreamed patch meets the approval bar for trademark approved downstream use is the fundamental communication breakdown here.

Looking at the bug history...
This was reported on March 5th upstream
This was approved upstream on April 1st.

So if I'm reading what you saying correctly...from Mozilla's POV... this would have been acceptable under trademark policy to use the patch downstream as of the approval on April 1st?

-jef

Fedora, Mozilla, and trademarks

Posted Apr 28, 2010 22:59 UTC (Wed) by roc (subscriber, #30627) [Link] (1 responses)

Well, we might not want to say "any patch on trunk can be backported to any branch, no problem" because that can often have undesirable side effects.

But I think we should certainly say that any patch approved to land on a particular branch (but not yet landed there) can be applied by anyone packaging that branch. That alone wouldn't have helped here since there was an extra problem --- the branch approval request was ignored for four weeks :-(. That was simply a mistake, people should never sit on requests for long. But if you read the Mozilla bug --- https://bugzilla.mozilla.org/show_bug.cgi?id=550455 --- there's not much communication from Fedora there. Just one comment on the 23rd saying this matters to Fedora, and the patch was branch-approved five days later (which is definitely too long, but not outrageous IMHO). A more detailed hurry-up nag, especially if it happened earlier, would have really helped.

So what should have happened IMHO is as soon as the Fedora packagers realized they needed this patch, they should have commented in the bug "our users are hurting and we need branch approval for this patch NOW NOW NOW", that approval should have been granted immediately, and then via a previously clearly-communicated blanket approval to use such patches with our trademarks, Fedora would have packaged the fix.

Fedora, Mozilla, and trademarks

Posted Apr 28, 2010 23:22 UTC (Wed) by jspaleta (subscriber, #50639) [Link]

"Well, we might not want to say "any patch on trunk can be backported to any branch, no problem" because that can often have undesirable side effects."

Granted, the question is how can what is allowed be more obvious and more explicit so its less likely someone will be waiting for approval mistakenly?

As for the unnecessary delay before it was made clear that this was a wide impactor. I'm sure an invitation for people to nag in upstream tickets more won't go...unrewarded. Something for the day-to-day Fedora Mozilla package maintainers and triagers and possible FESCO to be aware of.

-jef

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 4:04 UTC (Thu) by roc (subscriber, #30627) [Link] (10 responses)

I am informed that we have libffi changes that need to go upstream before we can use "system libffi". That is being worked on. Also, someone has submitted system-libffi patches and they are being reviewed and will be merged. So the libffi problem, at least, is being resolved.

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 5:51 UTC (Thu) by pabs (subscriber, #43278) [Link] (8 responses)

Shouldn't you be getting these changes merged upstream first and once they have been merged and released, then start depending on the version that includes them?

Autoconf is for this exact needs...

Posted Apr 29, 2010 6:15 UTC (Thu) by khim (subscriber, #9252) [Link]

These two processes can go in any order. Correctly written autoconf will just refuse to use library without changes so till upstream will include fix the autoconf will refuse all upstream releases.

But of course it's good idea to wait till upstream includes the patch at least in VCS: this will mean there WILL be supported release in the future...

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 8:05 UTC (Thu) by roc (subscriber, #30627) [Link] (6 responses)

Getting changes merged upstream first would slow down development a lot.

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 11:49 UTC (Thu) by ewan (guest, #5533) [Link] (1 responses)

Fedora has a strong policy of 'upstream first' and, to say the least, it doesn't generally get people complaining that the pace of change is too slow.

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 23:48 UTC (Thu) by roc (subscriber, #30627) [Link]

Fedora is not building a large application on top of those upstream libraries that needs features added to those libraries.

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 13:51 UTC (Thu) by ccurtis (guest, #49713) [Link] (3 responses)

Perhaps it would help if you posted a message like "our project is hurting and we need a release with this patch NOW NOW NOW" to their bug tracker.

Fedora, Mozilla, and trademarks

Posted May 2, 2010 3:31 UTC (Sun) by jwalden (guest, #41159) [Link] (2 responses)

Someone please correct me quickly if I'm wrong, but my understanding -- just from hearing people talking about it passing over lunch, or something like that, and this is the haze of a memory, so maybe I didn't hear correctly -- is that libffi isn't sufficiently strongly maintained to permit this.

I would also note (conveniently, one might say, but I believe accurately) that the browser market is a uniquely competitive one, as far as distribution-packaged software goes. What other distro package sees such widespread use, where time-to-market matters as much? (I'm speaking to the benefits of quick iteration through in-tree, lightly-patched third-party code now, not to the backport approval process.) I just skimmed my Applications menu in Fedora, and I can't see anything where new functionality and features have such high demand as for browsers.

Fedora, Mozilla, and trademarks

Posted May 3, 2010 20:12 UTC (Mon) by atgreen (guest, #33284) [Link] (1 responses)

I'm the upstream libffi author/maintainer.

Dan Witte, from Mozilla, has been terrific about feeding back libffi patches over the past few months. But from what I recall they mostly have to do with building libffi on Windows and OS/2. I don't remember anything that would prevent them from using the system libffi on Linux.

In any case, this sudden batch of contributions from Dan and Mozilla is appreciated.

Anthony Green

Fedora, Mozilla, and trademarks

Posted May 3, 2010 22:19 UTC (Mon) by jwalden (guest, #41159) [Link]

Ah, yeah, that's what I actually heard -- Windows support being the big thing. As I said, the memory was quite hazy. :-)

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 7:32 UTC (Thu) by glandium (guest, #46059) [Link]

> I am informed that we have libffi changes that need to go upstream before we can use "system libffi"

On the 1.9.2 branch, there are no upstream libffi changes required on linux to be able to use system libffi. Though one may wonder why js-ctypes is enabled on 1.9.2 at all, since it's only advertised for 1.9.3.

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 4:59 UTC (Thu) by tzafrir (subscriber, #11501) [Link] (2 responses)

I just wonder what would happen if libpng would have had the same branding requirements as Mozilla. Surely upstream knows best.

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 8:05 UTC (Thu) by roc (subscriber, #30627) [Link] (1 responses)

Then Mozilla wouldn't be able to use the libpng trademark. I think we'd deal.

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 8:44 UTC (Thu) by tzafrir (subscriber, #11501) [Link]

As someone reading this through IceWeasel, I fully agree with you that this would be the wise thing to do.

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 9:15 UTC (Thu) by niner (subscriber, #26151) [Link] (4 responses)

"MNG is a huge bloaty spec that has never been implemented by any mainstream browser."

If I am not mistaken it actually has. And ironically enough it was Mozilla that once upon a time had this support. It has been checked in 2000-06-12.

https://bugzilla.mozilla.org/show_bug.cgi?id=18574

Fedora, Mozilla, and trademarks

Posted Apr 29, 2010 23:51 UTC (Thu) by roc (subscriber, #30627) [Link] (2 responses)

OK OK. MNG is a huge bloaty spec that has never been shipped in any mainstream browser.

Fedora, Mozilla, and trademarks

Posted Apr 30, 2010 0:44 UTC (Fri) by foom (subscriber, #14868) [Link] (1 responses)

I couldn't care less about MNG, but...

According to https://bugzilla.mozilla.org/show_bug.cgi?id=18574#c653:
"It shipped in Mozilla 1.0 through 1.3, and Netscape 7.0, and had trivial usage on the web, and was
backed out before Mozilla 1.4 and Firebird 0.6.1."

Perhaps mozilla didn't really count as a mainstream browser back in 2001-2003? :)

Fedora, Mozilla, and trademarks

Posted Apr 30, 2010 4:47 UTC (Fri) by lambda (subscriber, #40735) [Link]

No, it didn't. If I remember correctly, Mozilla had maybe 1-2% market share back then, making it pretty far from mainstream. Those were pretty dark days, when there really was only one mainstream browser.

These days, there are 4 browsers I'd consider to be "mainstream" (IE, Firefox, Safari, and Chrome), and one, Opera, on the edge (in some countries, it has enough marketshare to be considered mainstream, and their mobile browser seems to be pretty popular).

Fedora, Mozilla, and trademarks

Posted Apr 30, 2010 0:34 UTC (Fri) by bronson (subscriber, #4806) [Link]

That was ten years ago! It seems pretty clear that MNG standard is stone dead.


Copyright © 2025, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds