Meeting:Packaging IRC log 20070109

From FedoraProject

Jump to: navigation, search
(09:00:42 AM) f13: hello all.
(09:01:06 AM) ***rdieter waves
(09:01:34 AM) lutter: hi guys
(09:01:39 AM) abadger1999: Hey guys,  meeting time
(09:01:46 AM) tibbs: Howdy.
(09:02:18 AM) rdieter: are we (still) spot-less?
(09:03:12 AM) rdieter: and atm, quorum-less as well? ):
(09:03:27 AM) tibbs: He seems to be marked "away" currently.
(09:03:28 AM) racor [n=rc040203]  entered the room.
(09:03:31 AM) rdieter: (wait, I can't count)...
(09:04:17 AM) rdieter: racor?
(09:04:24 AM) racor: yes?
(09:04:41 AM) f13: seems we've just got ratify's and writeups on the agenda, unless I can't read anymore
(09:04:51 AM) rdieter: ok, that makes 6 (if my basic math hasn't failed me)
(09:05:07 AM) ***f13 hopes to get through this quickly and still have time for lunch
(09:05:31 AM) ecik [n=ecik]  entered the room.
(09:05:35 AM) rdieter: I forgot to remove ratify from the icon_cache stuff...
(09:06:10 AM) rdieter: Any chance of passing the revised proposal?
(09:06:23 AM) lutter: tibbs: what's the status of Remi Collet's php stuff ?
(09:06:30 AM) scop [n=scop]  entered the room.
(09:06:57 AM) tibbs: I've heard no updates regarding what the core PHP maintainer decided to do.
(09:07:15 AM) tibbs: I know he asked for channel logs so he could read over the discussion.
(09:07:27 AM) lutter: that's joe orton ?
(09:07:31 AM) tibbs: Yes.
(09:07:56 AM) f13: rdieter: I"d like to get mclasen and co to look at the new proposal.  My fault for not driving that.
(09:07:59 AM) abadger1999: rdieter: What's the "desktop people" want to do WRT gtk-update-icon-cache?
(09:08:22 AM) rdieter: abadger1999: unclear what they want (though I'd argue it shouldn't matter, but that's just me...)
(09:08:35 AM) rdieter: ok, table icon_cache.
(09:08:40 AM) abadger1999: f13: On a similar note -- any progress on getting one of the desktop team on the Packaging Committee?
(09:09:02 AM) lutter: tibbs: do you want to ping him ? (I could, too, but I am not as familiar with teh issue as you are)
(09:09:05 AM) f13: erm, I didn't know I was supposed to drive that.  I'll ask around.
(09:09:25 AM) rdieter: abadger1999:  (I didn't know that was a plan either), but it seems to be a good one.
(09:09:50 AM) tibbs: lutter: I need to find the bugzilla ticket.
(09:09:54 AM) abadger1999: I don't know if you were; I thought I remembered you saying that was the plan, though.
(09:09:59 AM) abadger1999: I could be confused :-)
(09:10:23 AM) rdieter: Since there was ml activity related recently, can we quickly revist the LibtoolArchives proposal?
(09:10:49 AM) rdieter: http://fedoraproject.org/wiki/PackagingDrafts/LibtoolArchives
(09:11:29 AM) f13: I'm not sure if we need them to be on the committee full time, or if we just need to be better about inviting / informing key people regarding a proposal so that they have a chance to comment on it before we vote.
(09:11:46 AM) rdieter: f13: +1, makes sense
(09:12:14 AM) f13: rdieter: for the libtoolarchives, I'd like to get jakub to look at it from our tools team.
(09:12:14 AM) rdieter: we should all try to ping interested/related maintainers on proposals affecting them.
(09:12:38 AM) rdieter: sigh, ok, table that too for now... (:
(09:12:43 AM) f13: I'm pinging jakub right now.
(09:13:22 AM) ***f13 is a bit gunshy after the iconcache thing :/
(09:13:30 AM) abadger1999: f13: That's a good point.  The desktop people sometimes seem off in their own land, though.  Having someone on the Committee that interacted with us every week could be a plus.
(09:13:39 AM) rdieter: f13: don't sweat-it, that was mostly my fault.
(09:13:47 AM) abadger1999: If it was something they wanted to do, rather than something they felt they had to do.
(09:14:24 AM) f13: abadger1999: yeah, I don't think I'd have much luck finding somebody who would _want_ to interact each week, especially if its on matters that don't concern the desktop team's packages.
(09:14:48 AM) f13: but I think a better effort to outreach to those concerned on specific issues would go a long way to helping
(09:14:59 AM) f13: we tend to get stuck in our own world too
(09:15:10 AM) abadger1999: f13: k.  Makes sense.
(09:15:13 AM) rdieter: fyi, I plan on writing a proposal for updating guildelines wrt .desktop files, particularly, proper usage of Name= vs. GenericName= , any initial feelings/comments?
(09:15:36 AM) rdieter: I'm sure somehow this, imo, no-brainer will end up being controversial as well.
(09:15:51 AM) f13: heh.
(09:15:56 AM) abadger1999: Sounds controversial.  We need someone from the fedora-gnome-desktop team to give input.
(09:15:56 AM) scop: we'll see, go for it ;)
(09:16:14 AM) abadger1999: Even though I think I'm going to be for what you write :-)
(09:16:27 AM) rdieter: abadger1999: I don't see why codifying that we should follow .desktop upstream standards should be controversial... but hey. (:
(09:16:31 AM) tibbs: I agree, that issue needs to be decided.
(09:16:44 AM) abadger1999: (Our Gnome Name/GenericName stuff doesn't make a lot of sense to me.)
(09:16:50 AM) scop: rdieter, use of StartupNotify needs attention too
(09:16:53 AM) f13: the controversial part is who defines the standard.
(09:17:12 AM) rdieter: f13: bah, Name vs GenericName is just common sense.
(09:17:28 AM) rdieter: It's just that, afaict, gtk/gnome doesn't (yet) use/support GenericName.
(09:17:35 AM) f13: heh, I'm not saying I have a problem with it, I'll wait until I see the proposal and do some research
(09:18:24 AM) rdieter: ok, my job to write it up now.  (I'll make each bit separate, so not to block StartupNotify on these other bits)
(09:19:12 AM) rdieter: If need be, we'll make it a SHOULD, not MUST, to not to ruffle feathers too much.
(09:19:33 AM) rdieter: any other topics?
(09:19:47 AM) XulChris: i have a quick question
(09:19:50 AM) tibbs: We should discuss what's to happen to our mailing list.
(09:19:56 AM) f13: XulChris: shoot
(09:20:42 AM) XulChris: im writing a guidelines for 3rd party repos page and what should a dist tag be for a 3rd party repo (or suggested use), some repos like livna use lvn# which is wrong and should say fc#, but is fc#.lvn okay?
(09:21:55 AM) tibbs: Hmm.
(09:22:04 AM) rdieter: 3rd parties can do pretty much what they want, but I'd argue: fc# is preferred, fc#.<repo_tag> if they insist on adding something (but not recommended)
(09:22:10 AM) f13: depends.  Are they planning on replacing any Core/Extras packages?
(09:22:16 AM) scop: I think this will generate much more noise than it's worth; personally I wouldn't even post any recommendations at all
(09:22:38 AM) f13: I'm in favor of .fc# or .fc#.repo
(09:23:09 AM) rdieter: FYI: .desktop spec wrt Name/GenericName: http://standards.freedesktop.org/desktop-entry-spec/desktop-entry-spec-0.9.8.html#recognized-keys
(09:23:10 AM) scop: ...and in particular, I would advice 3rd party repos to use something *other* than the disttags in use in Fedora
(09:23:15 AM) abadger1999: scop: +0.9
(09:23:34 AM) f13: scop: why would that be?
(09:23:37 AM) rdieter: scop: -1 (but that's just me)
(09:23:43 AM) XulChris: scop: how do you tell if the package is for fedora or epel?
(09:24:18 AM) scop: suffice to say that this is just a sample of the noise I meant
(09:24:19 AM) f13: scop: n/m, I don't really care right now (:
(09:24:35 AM) f13: if there is nothing else, I"d really like to go eat lunch
(09:24:47 AM) rdieter: f13: enjoy.
(09:24:59 AM) lutter: yeah, I think we're good
(09:25:25 AM) XulChris: so i guess ill just ask on the ml?
(09:25:28 AM) abadger1999: I'm going to talk with rdieter about .la's a little but I'll post anything valuable to the ml.
(09:25:36 AM) abadger1999: XulChris: what's the value?
(09:25:37 AM) rdieter: ok.
(09:25:49 AM) XulChris: abadger1999: correct usage of dist tag?
(09:26:16 AM) abadger1999: XulChris: As scop says, everyone's going to have an opinion and in the end we can't tell 3rd parties what to do and they can't tell us what to do.  What's the value to the end user.
(09:26:28 AM) scop: "correct" is impossible to enumerate
(09:26:33 AM) XulChris: abadger1999: to know what the dist tag is actually meant to be used for
(09:26:48 AM) XulChris: if it doest matter why do we even have the tag?
(09:26:52 AM) rdieter: XulChris:  .fc# (or .fc#.<repo>) for Fedora, el# for RHEL is what you/we should recommend.
(09:26:52 AM) scop: "...to be used for" *in Fedora*
(09:27:29 AM) racor: " ... to be used for for add-on packages", too.
(09:27:46 AM) XulChris: scop: right, so any 3rd party repo that adds on to fedora should also use it
(09:27:50 AM) scop: -1
(09:29:08 AM) abadger1999: rdieter: libtool... What about static libraries?  they can benefit from having .la files.
(09:29:35 AM) abadger1999: But they are more problematic than loadable modules.
(09:29:46 AM) rdieter: abadger1999: sure, if static libs are to be kept included anywhere (that use .la files), their .la files will probably need to be kept as well.
(09:29:51 AM) racor: scop: it's the only way to assure (future) proper integration into FE (Note: X.lvn# > X.fc#)
(09:30:16 AM) scop: racor, no it is not
(09:30:22 AM) scop: just bump X
(09:30:22 AM) rdieter: abadger1999: I can't remember if the -static lib proposal mentions that.
(09:30:52 AM) XulChris: wasnt there some issue with jpp tags?
(09:31:18 AM) racor: rpmver 1-1.lvn6 1-1.fc6
(09:31:22 AM) abadger1999: rdieter: Hmm.. doesn't seem to.
(09:31:24 AM) rdieter: http://fedoraproject.org/wiki/PackagingDrafts/StaticLinkage, nope, though it probably should mention it.
(09:31:25 AM) racor: 1-1.lvn6 is newer
(09:31:47 AM) abadger1999: rdieter: +1
(09:31:54 AM) scop: racor, just bump to 2.fc6 and be done with it
(09:32:08 AM) scop: 1-1.lvn6 < 1-2.fc6
(09:32:24 AM) racor: rdieter: -1. Not subject of my proposal
(09:32:24 AM) rdieter: tho, that's tangential to the libtool_archives issue.
(09:32:38 AM) racor: rdieter: right
(09:32:40 AM) Rathann [n=rathann]  entered the room.
(09:32:44 AM) XulChris: scop: but lvn6 doesnt tell me anything about which repo the packages should be used against
(09:32:48 AM) rdieter: racor: ok, I'll mention it with libtool archive then.
(09:33:06 AM) racor: rdieter: OK
(09:33:09 AM) abadger1999: rdieter: Hmm... I had an issue but I think the use of -static package will make that go away
(09:33:37 AM) scop: XulChris, shrug
(09:33:49 AM) XulChris: scop: which is the _purpose_ of the dist tag
(09:33:50 AM) racor: abadger1999: I haven't checked detail, but I guess so, too.
(09:33:53 AM) abadger1999: namely: If you build the loadable module when the -static is present you'll generate a .la that references that library's .la file.
(09:33:53 AM) scop: if you rely on the disttag to tell you that, you're in trouble anyway
(09:34:05 AM) XulChris: scop: why is that?
(09:34:06 AM) scop: XulChris, not everyone agrees
(09:34:08 AM) rdieter: abadger1999: yup.
(09:34:16 AM) XulChris: scop: thats why we need the guideline
(09:34:28 AM) scop: that won't change my -1
(09:34:31 AM) rdieter: abadger1999: tho, it shouldn't harm anything (in theory). (:
(09:35:08 AM) XulChris: scop: are you voting -1 for a technical reason or because it would mean work for livna?
(09:35:12 AM) racor: XulChris: Nope, this is not Fedora's business. It's a third party's business
(09:35:12 AM) abadger1999: rdieter: Won't the resulting loadale-module-package need to Require the library-static package?
(09:35:26 AM) rdieter: abadger1999: no.
(09:35:36 AM) XulChris: racor: actually it was decided in a previous fesco meeting to write such a guideline
(09:35:41 AM) abadger1999: Or does libltdl only use the .la requires information for building, not running the module?
(09:35:55 AM) rdieter: abadger1999: missing items from plugins/loadable modules doesn't(shouldn't) break anything
(09:36:09 AM) rdieter: abadger1999: the former, afaict.
(09:36:20 AM) abadger1999: k.  That seems fine then.
(09:36:24 AM) racor: XulChris: What? where?
(09:36:41 AM) XulChris: racor: ultimately its 3rd party business but that doesnt mean we can offer hints as to how things should work
(09:36:42 AM) rdieter: abadger1999: and kinda the latter, though that's in question, I thought it loaded simply what was available at the time.
(09:36:47 AM) XulChris: racor: couple weeks ago, check logs
(09:37:11 AM) XulChris: racor: or maybe 3 weeks ago from tomorrows meeting
(09:37:15 AM) racor: XulChris: I never saw FESCo inform us.
(09:37:18 AM) rdieter: abadger1999: but that kde bugzilla entry I referred to in the ml, the responder claimed libltdl doesn't grok those items at all for loadable-modules.
(09:37:27 AM) racor: XulChris: EOF
(09:37:32 AM) XulChris: racor: im informing you now, i didnt inform you last week because everyone was on vacation
(09:37:33 AM) racor: s/F/T
(09:37:48 AM) XulChris: ?
(09:37:53 AM) abadger1999: rdieter: Hmm.. This is an area I haven't tested.
(09:37:57 AM) XulChris: so fesco has to officially inform pc?
(09:38:09 AM) rdieter: abadger1999: so either way, no harm no foul (hopefully)
(09:38:18 AM) abadger1999: This bug or a different one? http://bugs.kde.org/show_bug.cgi?id=93359
(09:38:29 AM) scop: XulChris, -1 because my idea of what disttag is differs from yours
(09:38:45 AM) rdieter: http://bugs.kde.org/show_bug.cgi?id=139445
(09:39:03 AM) XulChris: scop: differs from mine or fedoras?  i dont have any opinion other than what fedora uses
(09:39:10 AM) tibbs: XulChris: I'm in the logs and I'm not sure where it was decided by FESCo that the PC should develop a guideline on this.
(09:39:33 AM) tibbs: In fact I see much discussion that indicates that we shouldn't try.
(09:39:47 AM) XulChris: tibbs: i mentioned a guideline would be a good idea and several fesco members agreed and with dist tag it was suggested i ask here
(09:40:45 AM) tibbs: Well, I'm reading the logs from Dec 28 where this was discussed and I see nothing referred to the packaging committee except discussion about whether the dist tag for the next release should be "fc7" or "f7".
(09:40:46 AM) XulChris: so if no one here cares ill just use rdieter's suggestion as it makes the most sense
(09:41:19 AM) XulChris: tibbs: actually that part of the log was not pasted
(09:41:22 AM) rdieter: XulChris: it's just a recommendation afterall, no one else is bound to actually abide by it.. (:
(09:41:28 AM) XulChris: rdieter: exactly
(09:41:29 AM) tibbs: I am reading my personal logs.
(09:41:34 AM) abadger1999: rdieter: k.  If it's an unimplemented feature I kinda hate to rely on it continuing to be unimplemented :-(
(09:41:42 AM) XulChris: tibbs: its at the end of the meeting, thl recommends i ask here
(09:42:31 AM) rdieter: abadger1999: my point being that the level of breakage stays approx the same either way.  We're not relying on anything. (:
(09:42:53 AM) tibbs: XulChris: You mean this:
(09:43:00 AM) tibbs: [Thu Dec 28 2006]  [12:40:41]  <thl>      bpepple, that probably a decision for the PC (sorry, once a
(09:43:00 AM) tibbs: gain I try to move work to the PC...)
(09:43:18 AM) XulChris: tibbs: not sure what did bpepple say?
(09:43:33 AM) tibbs: He's talking about whether the tag should be "fc7" or "f7".
(09:43:51 AM) scop: XulChris, to put it another way, we seem to have understood differently what exactly is the meaning of disttag in Fedora
(09:44:16 AM) XulChris: scop: by we do you mean you and fedora?
(09:44:33 AM) scop: XulChris, me and you
(09:44:44 AM) XulChris: im still not clear on what you think it should be used for
(09:44:56 AM) scop: right
(09:45:10 AM) XulChris: do you want to clarify your position?
(09:45:21 AM) XulChris: is your position it can be used for anything
(09:45:49 AM) XulChris: completely worthless tag that has no meaning?
(09:46:01 AM) XulChris: should be removed from rpm, etc...
(09:46:09 AM) scop: I don't agree with the "which repo the packages should be used against" part which you said is the purpose of it
(09:46:18 AM) abadger1999: rdieter: http://sourceware.org/autobook/autobook/autobook_170.html
(09:46:24 AM) scop: basically, a dirty hack to assist lazy packagers
(09:46:28 AM) abadger1999: That seems to contradict the unimplemented portion.
(09:46:58 AM) XulChris: scop: please clarify what you think the dist tag is to be used for
(09:47:02 AM) rdieter: good, so my initial guess/experience was correct, thanks.
(09:47:03 AM) abadger1999: But okay.  If the module still loads even if lt_dlopen fails to find the .la then things are still okay.
(09:47:36 AM) Rathann: XulChris: IMHO the dist tag shouldn't be used at all, at least not in the way we do it in fedora
(09:47:43 AM) scop: XulChris, read "Purpose of the Dist Tag" at http://fedoraproject.org/wiki/Packaging/DistTag
(09:47:51 AM) scop: I don't know how to put it more clearly
(09:47:51 AM) abadger1999: I'll do some tests to make sure.
(09:47:56 AM) rdieter: abadger1999: true, unfortunately, lots of kde modules have undefined symbols, and rely on that feature to work.
(09:48:23 AM) XulChris: scop: that says the "original" purpose meaning that the purpose has changed
(09:48:25 AM) rdieter: abadger1999: and it may be a nightmare to track down, fix all the breakage as a result.
(09:48:37 AM) Rathann: or rather, I mean that it shouldn't be used in %{release}
(09:48:51 AM) rdieter: abadger1999: especially if upstream is unwilling (luke-warm at best) to help fix it
(09:49:03 AM) abadger1999: rdieter: I mean -- if the loadable module as a .la works even if the .la's it depends on are not present.
(09:49:18 AM) XulChris: scop: but anyway isnt that the same as what im saying?
(09:49:19 AM) scop: XulChris, just read the two bullet points below that - I gather *that* is the new purpose, if any
(09:49:32 AM) XulChris: scop: the same spec can be used for fc or epel
(09:49:33 AM) abadger1999: Then you're right, there's no foul.
(09:49:40 AM) scop: XulChris, no, you're saying "which repo the packages should be used against"
(09:49:46 AM) XulChris: scop: so for fc packages we use fc, and for epel packages we use epel
(09:50:07 AM) XulChris: scop: thats the same thing
(09:50:15 AM) scop: again, I disagree
(09:50:29 AM) scop: ...and will stop this discussion here, sorry
(09:50:33 AM) rdieter: abadger1999: ok, that's not 100% true then, unfortunately.  One reason we need to avoid any/all extranous .la file references (esp in other .la files)
(09:50:36 AM) Rathann: what does it mean, "which repo the packages should be used against"?
(09:51:01 AM) XulChris: it means which distribution release the package was built with
(09:52:37 AM) Rathann: I wouldn't guess you meant that, but ok
(09:52:54 AM) tibbs: So note that the fedora-packaging list is supposed to go away in the future.
(09:53:01 AM) XulChris: well you wouldnt want to use lvn6 packages with epel3 for example
(09:53:11 AM) tibbs: Does anyone have issues with doing all PC business in fedora-devel.
(09:53:31 AM) Rathann: tibbs: I do, in fact
(09:53:52 AM) Rathann: -devel has more traffic
(09:54:06 AM) Rathann: and most discussions are unrelated to packaging
(09:54:09 AM) racor: tibbs: yes. fedora-packaging is one of the very few useful lists amongst this fedora-list inflation
(09:54:19 AM) scop: I think fedora-extras-list and fedora-devel should be merged, and packaging left alone
(09:54:30 AM) abadger1999: scop: +1
(09:54:40 AM) tibbs: Folks who object to this need to actually object to the proposals that are being posted.
(09:55:18 AM) tibbs: So far the objections were basically noted and not acted upon.
(09:55:18 AM) racor: merge devel+test
(09:55:35 AM) Rathann: tibbs: I don't see any mails in -packaging about that
(09:55:54 AM) tibbs: The proposals aren't being sent to -packaging.
(09:55:57 AM) Rathann: so where are they posted?
(09:56:09 AM) thl: Rathann, fab list
(09:56:12 AM) racor: tibbs: you mentioning it here, is the first time I've heard about it
(09:56:13 AM) Rathann: ...
(09:56:25 AM) ***scop doesn't read fab-list at the moment
(09:56:29 AM) tibbs: Do note that I'm objecting to it as well.
(09:56:49 AM) Rathann: let me get this straight, you want to delete -packaging and you're NOT telling its members that you want to do it?
(09:56:51 AM) abadger1999: tibbs: f13 mentioned something in passing about "virtually splitting a mailing list" so you could easily filter or get only a portion of the traffic or something.
(09:57:11 AM) racor: thl: neither do I - I don't read read-only lists
(09:57:14 AM) tibbs: Can I summarize things as "there is significant objection by many members of the packaging committee"?
(09:57:25 AM) abadger1999: I don't know what it is but I might be okay with a setup like that.
(09:57:27 AM) ***Rathann is not a member but does object
(09:57:34 AM) thl: racor, I will post it sooner or later to all effected list
(09:57:35 AM) racor: yes, thl and FAB should learn to communicate,
(09:57:45 AM) racor: or quit
(09:57:48 AM) tibbs: abadger1999: If you have to "virtually split" a list, you should just split it.
(09:57:58 AM) Rathann: exactly
(09:58:17 AM) thl: racor, that might happen quite soon ;)
(09:58:21 AM) tibbs: Channels and subject tags are just hacks around having multiple lists.
(09:58:21 AM) racor left the room (quit: "Leaving").
(09:58:23 AM) f13: well, mailman supports topics
(09:58:26 AM) f13: they're just a pain to manage
(09:58:32 AM) f13: and to teach new people about using them.
(09:59:18 AM) abadger1999: f13:  So what do topics get you that two lists do not?
(09:59:22 AM) Rathann: f13: why not make a list about all of fedora and let people "virtually split" it then...
(09:59:28 AM) Rathann: -_-
(10:00:09 AM) f13: abadger1999: you can easily turn up the volume without having to do a subscribe dance, still filter on list-post for consolidation, etc...
(10:00:12 AM) f13: actually
(10:00:35 AM) thl: btw, before I leave; I spot promised to bring the "Conflicts" proposal up in todays meeting as more and more package start using it
(10:01:15 AM) thl: and we need clear guidlines for it before we do the big review of core packages..
(10:01:31 AM) tibbs: Well, spot wasn't here today so hopefully he will bring it to the mailing list.
(10:03:26 AM) abadger1999: Hmm.. I see a problem with the Conflicts proposal already :-(
(10:03:47 AM) tibbs: I don't think it was ever updated after our initial discussions about it.
(10:03:50 AM) abadger1999: Recommending alternatives for binaries isn't what I'd call a best practive.
(10:04:04 AM) racor [n=rc040203]  entered the room.
(10:04:32 AM) abadger1999: (in general.  Sometimes it is good but sometimes it's inappropriate.)
(10:04:33 AM) thl: tibbs, could the committee please discuss it for now?
(10:04:41 AM) thl: e.g. today?
(10:05:02 AM) thl: s/for//
(10:05:59 AM) tibbs: We certainly wouldn't want to ratify it without the person driving it being around.
(10:06:46 AM) tibbs: Plus I'm never sure what to do about the formal header spot puts at the top of his packaging documents.
(10:08:12 AM) tibbs: I also recall an objection to renaming manpages.
(10:08:22 AM) tibbs: I though the recommended thing to do was put them in different sections.
(10:10:59 AM) tibbs: Hmm, well, sorry thl; I'm not sure anyone is still around to discuss this.
(10:11:15 AM) thl: seem so :-/
(10:11:49 AM) scop: well, I think placing them in different sections sounds kind of dangerous
(10:12:16 AM) tibbs: It's been done forever, though.
(10:12:47 AM) scop: dangerous in the sense that if something is moved to a clearly wrong section just to avoid file conflicts, it just can't be a recommended practice
(10:13:12 AM) racor: scop: It's largely useless without changing man.config
(10:13:12 AM) tibbs: The whole point of manpage sections is to avoid conflicts.
(10:13:24 AM) scop: of course
(10:13:27 AM) racor: renaming man suffixes is safe and easy
(10:13:59 AM) tibbs: What do you mean by "man suffixes"?
(10:14:07 AM) scop: I have a hole in my man knowledge right there
(10:14:16 AM) tibbs: ".3" versus ".3qt", for example?
(10:14:17 AM) racor: using #<suffix> instead of #
(10:14:23 AM) racor: tibbs: yes
(10:14:27 AM) scop: racor, and install to... where?
(10:14:32 AM) racor: to #
(10:14:33 AM) scop: both in /usr/share/man/man3?
(10:14:41 AM) racor: scop: yes
(10:14:42 AM) tibbs: That's what I'm considering to be different sections.
(10:14:48 AM) scop: so how do I access the .#<suffix> manpage?
(10:14:55 AM) racor: using man
(10:15:09 AM) scop: I'm puzzled
(10:15:14 AM) tibbs: Just look in /usr/share/man3 right now.
(10:15:30 AM) tibbs: You should have all kinds of different blah.3x pages there.
(10:15:46 AM) scop: yes, but let's say I have foo.3 and foo.3x there
(10:16:02 AM) scop: if I type "man 3 foo", I get foo.3, right?
(10:16:18 AM) scop: how do I get foo.3x?
(10:17:06 AM) scop: tibbs, FWIW, by different sections, I understood moving eg. man3/foo.3 to man4/foo.4
(10:17:07 AM) racor: scop: gimme a couple of seconds
(10:17:28 AM) tibbs: The thing after the dot is the section.
(10:17:52 AM) scop: but the X in /usr/share/man/manX is not?
(10:18:06 AM) tibbs: The directory is just a simple hash to keep the directory size down.
(10:18:16 AM) tibbs: You can put all of the manpages in one directory if you want to.
(10:18:41 AM) scop: tibbs, I don't follow
(10:18:46 AM) scop: example:
(10:18:54 AM) scop: # pwd
(10:18:55 AM) scop: /usr/share/man/man3
(10:18:59 AM) scop: # cp zlib.3.gz zlib.1.gz
(10:19:03 AM) scop: # man 1 zlib
(10:19:03 AM) scop: No entry for zlib in section 1 of the manual
(10:19:18 AM) abadger1999: $ man 3x addstr
(10:19:18 AM) abadger1999: No entry for addstr in section 3x of the manual
(10:19:23 AM) tibbs: It's a search optimization.
(10:19:58 AM) tibbs: There's a configuration bit that controls where it looks for what.
(10:20:05 AM) tibbs: MANSECT, I think.
(10:20:21 AM) tibbs: It also defines search order.
(10:20:25 AM) racor: scop: man -a
(10:20:55 AM) racor: returns all matching man-pages
(10:20:58 AM) scop: racor, ouch
(10:21:12 AM) racor: try yum install Inventor-devel Coin2-devel
(10:21:30 AM) racor: man -a 3 SoTransform
(10:21:36 AM) tibbs: I wonder if man will even look at sections not defined in MANSECT.
(10:21:48 AM) racor: scop: This is how man on ALL system behaves
(10:21:55 AM) racor: even outside linux
(10:22:32 AM) scop: ok
(10:22:49 AM) racor: IIRC, solaris even applied it by default for their OS man-pages
(10:22:51 AM) scop: anyway, I do think this has some serious potential for confusion
(10:23:07 AM) scop: but I don't have necessarily better ideas at the moment either
(10:31:07 AM) XulChris: scop: dribble is looking to livna for guidance, is there a technical reason dribble should use only drb# instead of fc#.drb, if you dont want to answer its okay but i would be curious to know if there is a technical reason why we should *only* use drb#
(10:32:29 AM) scop: I don't see anything wrong with fc#.drb
(10:32:33 AM) racor: XulChris: As I tried to express before, 3rd parties are free use what they want.
(10:32:50 AM) racor: scop: Neither do I
(10:33:29 AM) scop: what I would find a bit silly if a 3rd party repo would use just fc#
(10:33:40 AM) XulChris: racor: yes but using a dist tag that has meaning is better than one that has no meaning, so we should define clearly what the dist tag is used for and how it can be helpful for endusers as well (for example fc#.repo helps a human understand more about a package than just .repo#)
(10:34:45 AM) racor: scop: I do for my packages
(10:35:12 AM) scop: racor, you're free to do so, and I'm free to think it's a bit silly, and you're free to disagree ;)
(10:35:51 AM) racor: scop: I interpret %{?dist} as "distribution being targeted"
(10:36:19 AM) racor: scop: %{?dist}.<3rdparty> also seems reasonable to me
(10:36:40 AM) scop: I interpret it as "I'm lazy" ;)
(10:36:57 AM) thl is now known as thl_afk
(10:37:41 AM) racor: why? I target many distros, and don't have any use for anything else but a tag for the target distro
(10:38:14 AM) racor: because the packages install outside of "/usr" and can't conflict with the distro
(10:39:34 AM) scop: between completely different distros the issue is largely moot
(10:40:07 AM) racor: details: ftp://ftp.rtems.org/pub/rtems/linux/*/{redhat,fedora}/*
(10:40:15 AM) scop: between different versions of the same distro it's more or less the "I'm lazy and don't want to bump" case
(10:41:01 AM) XulChris: scop: having to bump is the only reason i can think of for only using fc# as the dist tag
(10:41:27 AM) XulChris: and fc#.repo conveys more info to an end-user
(10:41:42 AM) racor: XulChris: Exactly, it's the point behind %{dist}
(10:42:20 AM) XulChris: racor: its *one* of several points for using dist
(10:42:23 AM) scop: I agree that fc#.repo seems to have most benefits
(10:42:33 AM) XulChris: scop: me too
(10:43:23 AM) racor: scop: It will surprize you: yes, it has, if you are building a package folding into /usr and ....
(10:43:37 AM) racor: if %{dist} is mandatory for the base distro
(10:43:40 AM) scop: racor, note that I'm not fully implying lazy==bad in this case
(10:44:31 AM) scop: well, no distro I'm aware of has mandatory %{dist}
(10:44:38 AM) racor: in fedora it is not, therefore %{dist} is likely to clash with anything Fedora choose, so 3rd parties using $dist is dangerous
(10:45:43 AM) scop: and FWIW, I don't even think about upgrading base distro packages
(10:45:44 AM) racor: example: should livna ship xxx-1-2.lvn6 and RH decide to ship xxx-1-2, bad things will happen
(10:46:01 AM) racor: well, or FE ...
(10:46:12 AM) scop: depends
(10:46:44 AM) scop: some people will say that the above situation (xxx both in foo and bar repos) needs to be avoided
(10:46:56 AM) racor: yes, depends on whether Fedora recognises/knows about "3rd party"
(10:47:28 AM) racor: as Fedora you can't know about each and every 3rd party
(10:47:40 AM) scop: if upgrading a base distro package is the goal, there are certainly some changes to the package, otherwise you wouldn't be upgrading it
(10:47:59 AM) scop: and in that case, one should bump the EVR anyway, not rely on %{dist}
(10:48:06 AM) racor: scop: You misunderstood my example.
(10:48:36 AM) racor: I was referring to 3rd party shipping a packages, which at some point is being added to Fedora (FE)
(10:48:49 AM) scop: yes
(10:48:58 AM) scop: disttag is completely moot there, no?
(10:49:26 AM) XulChris: racor: right, when you add it to FE you should atleast make a release bump anyway saying you are moving package from repoX to FE
(10:49:27 AM) racor: that's the point: disttag is mostly pointless for 3rd parties
(10:49:44 AM) XulChris: racor: from that perspective yes, but it has other uses
(10:49:59 AM) racor: but as it's handy to users, and because the risks are pretty low ...
(10:50:13 AM) racor: XulChris: ACK
(10:50:45 AM) racor: sorry, boys I've got to quit
(10:50:48 AM) racor left the room (quit: "Leaving").
(10:50:52 AM) scop: me too, actually