Packaging:Minutes/20080826

Fedora Packaging Committee Meeting 2008-08-26

Members Present

• Denis Leroy (delero)
• Dominik Mierzejewski (Rathann|work)
• Hans deG oede (hansg)
• Jason Tibbitts (tibbs)
• Ralf Corsepius (racor)
• Rex Dieter (rdieter)
• Tom Callaway (spot)
• Toshio Kuratomi (abadger1999)

Summary

• Approved the draft Haskell packaging guidelines: PackagingDrafts/Haskell
• Approved the draft Lisp packaging guidelines: PackagingDrafts/Lisp
• Discussed two submissions related to the bundling of fonts: PackagingDrafts/Packaging_font_bundles and PackagingDrafts/No_bundling_of_fonts_in_other_packages.
  • Neither was approved as-is, but as a result of the discussion, FPC approved one general guideline addition and two related documents:
    • "Fedora packages should make every effort to avoid having multiple, separate, upstream projects bundled together in a single package."
    • TomCallaway/Packaging_Font_Bundles2
    • TomCallaway/No_bundling_of_fonts_in_other_packages2

IRC Logs

*** tibbs sets the channel topic to "Packaging Committee Meeting".		12:00
* Rathann present		12:00
* tibbs here		12:00
* spot is here		12:01
racor	i am here, have ca. 15 mins time.	12:01
spot	racor: thanks for coming	12:02
tibbs	rdieter, abadger1999: ping	12:02
rdieter	here	12:02
tibbs	Anyone else I missed?	12:02
--> delero has joined this channel (n=denis@AMontsouris-156-1-79-180.w90-24.abo.wanadoo.fr).		12:03
abadger1999	here	12:03
delero	here	12:03
tibbs	That's seven.	12:03
spot	we're missing hans and Xavier	12:04
spot	but hey, quorum. :)	12:04
abadger1999	Woo hoo :-)	12:04
spot	okay, first order of business, the drafts that i sent around via email	12:04
spot	some of you voted over email, thanks.	12:04
spot	however, some of you did not	12:04
tibbs	I saw five sets of votes including my own.	12:05
spot	yep.	12:05
* Rathann wonders why the members list link on FPC wikipage points nowhere		12:05
abadger1999	Haskell +1, Lisp: +1, fonts.... I'd like to know if the fonts sig will enforce those if we vote 0.	12:05
spot	delero: would you like to vote?	12:06
spot	we have votes for everyone else	12:06
Rathann	abadger1999: or at least how much work it is to enforce that	12:06
* rdieter just sent email minutes ago, in short, I +1'd all of them (including fonts).		12:06
tibbs	Rathann: Probably more damage from the conversion. I keep cleaning things up but there's always something else.	12:06
delero	i went over them, +1 for me for all 3	12:06
tibbs	There were four.	12:07
Rathann	but two were about fonts and related	12:07
tibbs	True. But two separate proposals.	12:07
Rathann	yup	12:07
spot	okay, haskell and lisp clearly pass	12:08
Rathann	which is why I should add my vote for font bundles: 0	12:08
spot	On the Lisp draft, the following comments were made:	12:08
abadger1999	I like both the font proposals. If the fonts-sig is going to enforce them anyways 9as part of a SIG best practice) then I have even more reason to vote +1.	12:08
spot	Needs adding an ASDF system definition file template (or link to syntax).	12:08
spot	needs to add an empty %build to his spec template	12:09
tibbs	I think the link at the bottom should suffice.	12:09
delero	i'm ok with the font bundling proposal as well, +1 from me	12:09
spot	my concern around the font bundling proposal is that they were drafted specifically to prevent texlive	12:09
abadger1999	spot: Err... specifically in response to texlive.	12:09
spot	yes, rather.	12:10
spot	while i think that texlive is a clear exception to that guideline	12:10
Rathann	I wonder how much work it would be for texlive packager to adapt texlive to follow that guidline	12:10
tibbs	Well, texlive does need cleanup.	12:10
racor	my concern is that fonts are being bundled with other sources, whatever the font sig wants doesn't change much about it	12:11
tibbs	But it's going to have to evolve in that direction.	12:11
rdieter	grandfather'd exception for texlive, +1 (that doesn't mean that efforts to fix it shouldn't happen)	12:11
Rathann	if it's doable in reasonable time, then I'd vote +1 on both font proposals	12:11
tibbs	I think the "no bundling of fonts" proposal is simply unworkable.	12:11
tibbs	If I have a game or something that has a simple bitmap font in its own internal format.	12:12
Rathann	hmm	12:12
Rathann	good point	12:12
tibbs	It's a "font" according to the guideline, but I doubt there's going to be any call to split it.	12:12
spot	ok, lets vote on the font bundles proposal first (with an exception for texlive for the time being)	12:12
racor	-1	12:13
tibbs	-1	12:13
rdieter	+1	12:13
spot	this is http://fedoraproject.org/wiki/PackagingDrafts/Packaging_font_bundles	12:13
abadger1999	+1	12:13
tibbs	I also have to wonder why fonts are special here.	12:14
Rathann	tibbs: because they can be used by other apps?	12:14
tibbs	The arguments work for more than just fonts.	12:14
spot	well, to be fair, we don't generally permit multiple software items from multiple sources to live in the same srpm	12:15
abadger1999	tibbs: That's true. I could go for a more general rule.	12:15
racor	Rathann: Many files can be used by other apps (shared libs, images, sound, movies ...)	12:15
rdieter	tibbs: good point, there's currently a best-practice/unwritten rule already about separate sources => separate pkgs.	12:15
abadger1999	But that doesn't eliminate my +1 for the subset :-)	12:15
tibbs	I don't know if everyone saw my suggested alternative.	12:15
rdieter	tibbs: please refresh our memory.	12:15
tibbs	I would	12:16
tibbs	consider it it were distilled to a simple strong suggestion that separate	12:16
tibbs	upstream projects not be bundled together in the same package. I	12:16
tibbs	believe that's an unwritten rule already.	12:16
spot	tibbs: "a simple strong suggestion that separate upstream projects not be bundled together in the same package."	12:16
tibbs	But I don't want to derail the current vote. I can write a proposal later.	12:16
rdieter	I can totally support that.	12:16
spot	i'm much more in favor of that	12:16
Rathann	0 unless a list of font formats to which this guideline is applicable is supplied (+1 then)	12:16
abadger1999	tibbs: I'd love to have that written down rather than unwritten.	12:16
delero	strictly enforcing it on existing packages is going to be tough, especially on dormant projects where the fedora packager has become the ad-hoc maintainer	12:17
abadger1999	Although textlive, for instance, has already brought issues up wrt that :-(	12:17
Rathann	tibbs has a good point about bundled fonts that could not be used system-wide	12:17
tibbs	Enforcing anything on existing packages has always proven difficult.	12:17
delero	pstoedit is an example, it ships with an old bitmap font	12:17
abadger1999	as texlive is an upstream but is also a conglomeration of other upstreams, how does the rule apply?	12:17
abadger1999	I don't need that answered now, just saying that question has already been raised.	12:18
Rathann	also, it doesn't always make sense to make some obscure fonts (symbol fonts, incomplete fonts) visible system-wide	12:18
spot	I think that we should say something like "Fedora packages should make every effort to avoid having multiple, separate, upstream projects bundled together in a single package."	12:18
rdieter	I'd say let's ask for the current bundles draft to strike paragraph 1, pending a more general soon-to-come guideline.	12:19
Rathann	so while I agree with the guideline in the spirit, it needs to allow for common sense and not be as strict	12:19
abadger1999	I just got a call. I have an electrician coming out to the house and will have to leave when he gets here.	12:19
delero	Rathann: +1	12:19
rdieter	Rathann: I read it that way already, it says SHOULD	12:19
Rathann	ah	12:20
Rathann	right, the last paragraph of http://fedoraproject.org/wiki/PackagingDrafts/No_bundling_of_fonts_in_other_packages clears it up a bit	12:20
Rathann	but I'd limit it to what I suggested above	12:21
racor	i think paragraphs 2+3 should be striked. They attempt to special case something which isn't a special case.	12:21
Rathann	i.e. not only general-purpose formats but general usability	12:21
tibbs	rdieter: The Packaging font bundles proposal doesn't seem to say SHOULD. Lots of MUSTs there.	12:21
rdieter	I'm reading http://fedoraproject.org/wiki/PackagingDrafts/Packaging_font_bundles , which includes MUST only in paragraph 1, which I think we've all agreed needs to be stricken anyway	12:22
spot	with two -1, and one 0, there is no way that "Packaging_font_bundles" can pass	12:22
rdieter	the only must remaining is the licensing bit, but maybe that's not required hee	12:22
tibbs	Email from Xavier: he'll be a bit late.	12:22
Rathann	does "packaged separately" mean a totally separate package or can it be a subpackage?	12:22
Rathann	hm looks like the former	12:23
rdieter	Rathann: I read that as either	12:23
tibbs	I read it as "totally separate package".	12:23
racor	source or binary package?	12:23
rdieter	sorry. context matters. ignore me	12:23
tibbs	Otherwise arguments about separate upstream release cycles and such make no sense.	12:23
racor	demanding a separate source package is silly	12:23
spot	can we vote on adding "Fedora packages should make every effort to avoid having multiple, separate, upstream projects bundled together in a single package." to the main guidelines?	12:23
rdieter	spot: +1 to that	12:23
Rathann	spot: +1	12:23
delero	spot: +1	12:23
tibbs	Will we need to discuss exemptions?	12:24
racor	-1, superflous not of any importance	12:24
tibbs	Circular dependencies was always an interesting one to me.	12:24
tibbs	+1	12:24
spot	+1 from me	12:24
tibbs	racor: What existing guideline does this duplicate?	12:24
Rathann	it may be common sense, but common sense is sometimes most difficult to follow ;)	12:25
spot	with a +5, it passes	12:25
spot	now, it seems like we might be able to reword the first paragraph of Packaging_font_bundles to make it more sensible	12:25
abadger1999	Would it be better to phrase it as a MUST? ie: any package which bundles multiple separate upstream projects MUST justify that decision?	12:26
racor	tibbs: this sentence is a waste of text - whether this sentence is presence or not doesn't change anything	12:26
tibbs	I don't follow your argument.	12:26
racor	upstreams don't care about what we decide	12:26
racor	to new-comer packagers this text is not helpful	12:27
racor	it's just bloat	12:27
tibbs	It answers a question that has been asked of me several times already.	12:27
tibbs	I guess I could simply continue to answer as I wish, but I'd rather have an actual guideline. Which it seems we'll have.	12:27
* rdieter is confused now, this guideline only describes downstream packaging, not much to do with upstreams at all		12:28
racor	bring this to attention of major upstream projects - You'll be laughed at.	12:28
Rathann	racor: some upstreams start caring when we explain it to them	12:28
spot	okay, so, if we replace the first paragraph of Packaging_font_bundles to "As noted in the Packaging Guidelines, Fedora packages should make every effort to avoid having multiple, separate, upstream projects bundled together in a single package. This applies equally to font packages."	12:28
spot	then, leave the rest as is... i think that makes it more reasonable	12:29
racor	Rathann: Do you care about SuSE, Debian, Ubuntu, Gentoo packaging desires in packages you are upstream? I don't.	12:29
rdieter	I'm still not sure about the "each bundled font set ends up in a different mono-licensed sub-package", that seems to be itching for a generalized rule too, no?	12:29
Rathann	racor: nobody laughed at me when I started packaging inchi separately from openbabel and submitted patches to fix building with external inchi	12:29
racor	try glibc, try gcc, ...	12:30
spot	rdieter: i think it makes some sense to do it that way for fonts specifically, especially if other applications want to rely on a single font	12:30
racor	anyway, i've got to quit now, sorry.	12:30
spot	they wouldn't need to Requires: foo-superfonts-dump, they could Requires: foo-superfonts-myfont	12:31
rdieter	umm... do apps really need to care about the fonts used licensing-wise?	12:32
rdieter	if so, doesn't that get very scary, very fast?	12:32
spot	rdieter: generally, no, but it does make for a reasonable divisor	12:32
Rathann	but licensing is a good reason for splitting packages in general	12:33
Rathann	i.e. foo licensed under GPL and foo-libs under LGPL	12:33
Rathann	assuming that's what upstream does	12:33
Rathann	or foo-someplugin under another license (say, BSD)	12:34
abadger1999	You have my +1 to this.	12:34
abadger1999	Electrician is here, gotta run.	12:34
Rathann	I wouldn't put so much stress on packaging fonts separately due to licensing issues, it is a more general thing	12:35
spot	take a look at this:	12:35
spot	https://fedoraproject.org/wiki/TomCallaway/Packaging_Font_Bundles2	12:35
tibbs	The first MUST there implies that this guidelines is stronger than the general one.	12:36
Rathann	doesn't say why, though	12:36
tibbs	Is that still the intention, or is this just supposed to be a clarification of the other guideline as it applies to fonts?	12:36
delero	this means ONE font per subpackage ?	12:36
Rathann	delero: one font family	12:36
spot	delero: font family	12:37
* rdieter likes that draft more. happy happy		12:37
spot	i think this is a clarfication for fonts	12:37
Rathann	well if it's phrased that way, it becomes redundant	12:38
Rathann	the first paragraph	12:38
Rathann	but +1 too	12:38
delero	I assume this would apply to a package like gnuplot, which ships postscripts fonts	12:38
rdieter	not this draft, afaict, the other one... maybe. :)	12:39
Rathann	delero: but does it make sense to use them outside gnuplot?	12:39
spot	well, should we take a vote on https://fedoraproject.org/wiki/TomCallaway/Packaging_Font_Bundles2 ?	12:40
delero	Rathann: unlikely	12:40
Rathann	I wouldn't want to force any package to split their fonts if it doesn't make sense to use them outside their apps	12:40
spot	keep in mind that all reasonable exceptions are okay.	12:40
spot	(as always)	12:40
rdieter	spot: +1 P_F_B2 draft	12:40
delero	spot: +1 on v2 draft	12:40
spot	+1 from me	12:40
Rathann	+1 on v2 draft	12:40
spot	abadger1999: gave us a +1 before he left... tibbs?	12:41
tibbs	I guess I don't really understand why it mandates a split by license, but I don't have any real problems with it.	12:42
tibbs	+1	12:42
spot	tibbs: it doesn't mandate that anymore	12:42
spot	i changed it to read "make sure each bundled font set ends up in a different, appropriately licensed sub-package. "	12:42
Rathann	rather, splitting by license is a general "should, if makes sense"	12:42
tibbs	I've been reloading but I don't see the change.	12:42
Rathann	not only with fonts	12:42
spot	tibbs: https://fedoraproject.org/wiki/TomCallaway/Packaging_Font_Bundles2 ?	12:42
tibbs	I'm still seeing "but he MUST make sure each bundled font..."	12:43
spot	note that i made a copy with changes for v2	12:43
tibbs	Yeah, I'm looking at v2.	12:43
spot	If upstream refuses the packager MAY base a single src.rpm on the collection archive, but he MUST make sure each bundled font set ends up in a different, appropriately licensed sub-package.	12:43
spot	old version said "If upstream refuses the packager MAY base a single src.rpm on the collection archive, but he MUST make sure each bundled font set ends up in a different mono-licensed sub-package."	12:44
spot	anyways, thats +6	12:44
spot	i can't think of a quick way to reword No_bundling_of_fonts_in_other_packages in such a way that it would be acceptable	12:45
spot	well, i take that back	12:45
spot	maybe if we changed item 1 to	12:45
spot	1. any package that makes use of fonts should strongly consider packaging them in a separate sub-package, if they have any value outside of the package	12:46
rdieter	item 2 isn't really a MUST, just a pointer	12:46
Rathann	spot: I'm fine with a MUST in your modified version even	12:47
spot	yeah, but if it makes people think about font licenses, i'm not opposed to it	12:47
rdieter	spot: your version of 1 is a lot better, likey likey	12:48
--> hansg has joined this channel (n=hans@ip32-174-211-87.adsl2.static.versatel.nl).		12:48
hansg	Hi all	12:48
Rathann	also if such font has value outside the application, maybe ask upstream to publish font source separately?	12:48
hansg	I just saw spot's invitation	12:48
hansg	any votes needed from me, or did we already have the quorum?	12:49
spot	hansg: you're not too late, we're just going through the last pending draft	12:49
Rathann	hansg: we're discussing http://fedoraproject.org/wiki/PackagingDrafts/No_bundling_of_fonts_in_other_packages	12:49
hansg	-1	12:49
spot	hansg: we're trying to fix it. :)	12:49
Rathann	<spot> maybe if we changed item 1 to	12:49
Rathann	<spot> 1. any package that makes use of fonts should strongly consider packaging them in a separate sub-package, if they have any value outside of the package	12:49
hansg	For reasons already mentioned	12:49
Rathann	well? any comments on my suggestion?	12:50
spot	https://fedoraproject.org/wiki/TomCallaway/No_bundling_of_fonts_in_other_packages2	12:50
spot	take a look at that	12:50
spot	Rathann: i'm on the fence as to whether it should be a "SHOULD" or a "MUST"	12:51
spot	i can see both sides of that argument	12:51
spot	(i'm leaning towards a must, as a sub-package)	12:51
Rathann	spot: no, I mean the other suggestion ;)	12:52
Rathann	I said I was fine with either should or must here	12:52
spot	oh yes, that is good, i'll add it	12:52
hansg	Hmm, just read Spot's draft I dunno what to think of this	12:52
Rathann	hansg: it's just the 1st point that's different	12:52
Rathann	I think that was the main contention	12:53
hansg	All in all it seems well balanced between making clear that generic fonts must be packaged separately and that specials could be bundled	12:53
hansg	I would like to see some language in here about how this all applies only to fonts in font format.	12:53
Rathann	"fonts in font format"?	12:54
tibbs	Think back to my earlier question.	12:54
Rathann	hansg: the last paragraph is not enough?	12:54
hansg	Games often package fonts as just a bmp which when you lay a 64x64 grid over it you get each letter	12:54
Rathann	ah	12:54
Rathann	then the first solves it	12:54
hansg	or on XxX format for that mayyer	12:54
spot	okay, i added Rathann's suggestion: https://fedoraproject.org/wiki/TomCallaway/No_bundling_of_fonts_in_other_packages2	12:54
spot	hansg: if i made it say "bundled font files"	12:55
spot	would that be more appropriate?	12:55
spot	or "bundled fonts (in font format)"	12:55
hansg	I'm not sure about the "any value outside of the package " wording, that is a bit vagu	12:55
hansg	vague I mean	12:56
hansg	Spot, +1 for "bundled font files"	12:56
hansg	That is better IMHO	12:56
hansg	And maybe we should replace the "any value outside of the package " wording by a list of formats and a MUST be in a subpackage if in one of these formats	12:56
Rathann	hansg: it's not just formats	12:57
Rathann	it's also the usability	12:57
hansg	For example if a game has some special font created for it in ttf, it would be good to put it in a sub package so that it can be used more general	12:57
hansg	(assuming the licensing is ok)	12:57
hansg	Rathann, explain	12:57
Rathann	it doesn't make sense if such font contains just a limited subset of characters or symbols which is usable only in that game	12:57
rdieter	hansg: but "good" for who? if no one can/will ever use it?	12:58
Rathann	hence it also needs to be usable outside the original application	12:58
hansg	Rathann, ah yes	12:58
hansg	rdieter, will never use it just a matter of advertising, can never user it is another story	12:58
spot	how about something like "especially if the font is in a standardized format, and contains a set of characters or symbols which are useful for other packages."	12:58
hansg	Ok, lets stick with the "any value outside of the package "	12:59
hansg	spot +1	12:59
hansg	(although that violates the less is more principle)	12:59
spot	it does, and i think the "any value outside of the package" is actually broader.	12:59
hansg	I'm fine with keeping just "any value outside of the package "	13:00
* Rathann is fine with either		13:00
Rathann	obviously some people need explaining the spirit of the rule ;)	13:00
Rathann	so more precise language won't hurt IMHO	13:01
spot	how about this	13:01
spot	https://fedoraproject.org/wiki/TomCallaway/No_bundling_of_fonts_in_other_packages2	13:01
spot	i added it as a clarifier	13:01
tibbs	Seems OK to me.	13:02
Rathann	yup	13:02
delero	excellent	13:02
tibbs	I wonder whether we're still within the spirit of the original draft which was submitted.	13:02
spot	well, we'll hear back for sure if we're not.	13:02
spot	i think we are	13:02
hansg	spot, good, about font source, its that mandatory, or does it depend on the font license?	13:02
Rathann	I think so too	13:02
Rathann	we just added some common sense ;)	13:02
spot	hansg: not sure i follow	13:03
Rathann	hansg: you mean the point about asking upstream to publish font source separately?	13:03
Rathann	it's not mandatory	13:03
hansg	Well if someone designed a font for project X and gave project X just the .ttf file and a license to do whatever they want with the ttf, there will be no font source	13:03
spot	how about i change that to just "font files"	13:04
spot	not "font source"	13:04
spot	to eliminate confusion	13:04
Rathann	yes	13:04
Rathann	that's what I meant	13:04
hansg	spot, hmm yes and no	13:04
hansg	We do want the preferable format for modification when available	13:04
<-- delero has left this channel.		13:05
spot	hansg: yes, but thats a licensing issue	13:05
Rathann	the idea is that generally useful fonts should migrate to separate packages alltogether	13:05
Rathann	hence my suggestion to ask upstream to publish font files	13:05
hansg	spot again yes and no, it can be that we don't need the font "source" from a license pov, that doesn't mean we don't want it if available	13:05
spot	by not specifying, we can safely assume we want source	13:06
hansg	true	13:06
spot	https://fedoraproject.org/wiki/TomCallaway/No_bundling_of_fonts_in_other_packages2	13:06
spot	it now says "font files" on that line	13:06
spot	guys, i need to dash off to the board meeting	13:06
spot	can we vote on this quickly?	13:06
Rathann	hansg: but isn't that is already in font packaging guidelines?	13:06
rdieter	spot: +1 to v2	13:07
spot	+1 from me	13:07
hansg	+1	13:07
Rathann	hansg: "Fonts SHOULD be built from source whenever upstream provides them in a source format"	13:07
tibbs	+1	13:07
Rathann	+1 from me	13:07
spot	do we have quorum? delero dropped...	13:07
hansg	Rathann, we should add: "if upstream does not provides them in a source format, the packager should contact upstream and ask them to provide source if possible"	13:07
spot	wait, that is +5	13:07
spot	ok, it passes.	13:07
hansg	yes thats enouh, right??	13:08
spot	and with that i have to go to the board call	13:08
Rathann	hansg: that is fine by me	13:08
spot	thanks guys	13:08
Rathann	thanks spot ;)	13:08
* hansg wonders why my typing is even lousier then normal		13:08
spot	i'll update the todo page this afternoon	13:08
tibbs	I will try to get minutes out today; with the FESCo move, we have no way to make their 24 hour deadline.	13:09
hansg	well that was short (for me) when is the next meeting?	13:12
Rathann	hansg: we should discuss the possible times and choose one that works well for everyone if possible, current one was difficult for me until recently	13:17
hansg	Didn't we try that by using a wiki page were we all wrote down what worked, and then failed?	13:18
Rathann	and I won't know until next week if it continues to be	13:18
tibbs	The problem seems to be that there's always some kind of conflict.	13:18
abadger1999	tibbs: FTR, I'm +1 to spots revised fonts guideline.	13:33