gaim ebuild tries to install nss and nspr libraries when 'mozilla' is in USE flags and those libraries are already installed as part of mozilla. gaim seems to compile and work fine with mozilla NSS/NSPR libraries.
Created attachment 42822 [details, diff] use mozilla nss/nspr libs if 'mozilla' in USE flags
Created attachment 42823 [details, diff] use mozilla nss/nspr libs if 'mozilla' in USE flags
sorry, created the wrong diff the first time. second patch is the good one.
I'm not sure I agree. It's better to track the nss/nspr libraries that are in portage, as we have a better idea about the API/ABI changes going on in these than we do in random Mozilla releases.
Aren't the USE flags for exactly this kind of choice? Please provide the choice for users who have 'mozilla' flag defined. Anybody who has mozilla and wants to build it with native nss, can still do 'USE=-mozilla emerge gaim' after this patch. But users who want to use mozilla build, they don't have a choice at this moment.
I used to have it try this. But then there was the issue of do we do mozilla, or mozilla-firefox, or what. There are a few packages (thunderbird maybe) that provide NSS and NSPR. Rather than deal with the headache, we (me and Brad the mozilla master) agreed that just depending on standalone libs would be best. Does the 'mozilla' USE flag imply that you want the full Mozilla, even though firefox would be sufficient in this case?
we stop at mozilla and don't search for firefox or thunderbird. the reason being that mozilla drives other browser(epiphany, galeon etc.) and package installs as well and hence is more likely to be present if USE contains 'mozilla'. And of course, checking for firefox/thunderbird will involve more code changes in the ebuild. if you think that we should check for firefox/thunderbird as well for USE=mozilla, I can update the patch for that. but as I said, I think we should stop at mozilla. user has no business setting 'mozilla' USE flag if he is not ready to install mozilla as dependency for other packages which require mozilla headers/libs.
gaim's msn plugin doesn't work if /usr/lib/nss isn't in ld.so.conf, something that should be in env.d i guess?
The /usr/lib/nss issue is another bug currently open for dev-libs/nss.
I'm going to close this and decline the change. As I've said we (both gaim and evolution and some others) hashed this problem around and all decided it was just easiest to depend on the rather small standalone libs anyway.
I still don't understand this. If a user has 'mozilla' in USE flags, doesn't it mean that he has mozilla installed for some purpose(may be other than gaim)? All I am asking for is an option for such users in the gaim ebuilds. It doesn't break anything. If they can't compile it with mozilla headers(which I don't see any reason for, just some ABI paranoia. I have always compiled gaim with mozilla headers and linked with mozilla nss/nspr libs, ever since I found gaim), those users have the option with "USE=-mozilla emerge gaim". Users who don't have mozilla in their USE, are not affected at all. I am not saying 'don't depend on small independent nss/nspr libs'. But depend on them conditionally. That's what ebuilds are there for, right? Somebody please give me a compelling reason for not including this request. Please update the bug database directly rather than have discussion among yourselves and then just closing the bug.
Both mozilla, firefox, and nss put their lib paths in /etc/ld.so.conf so I no longer need to specify their locations. I'm going to do my old dep like this: !gnutls? ( || ( >=dev-libs/nss-3.9.2-r2 net-www/mozilla-firefox net-www/mozilla ) ) I don't need the mozilla use flag. This will detect if you have mozilla firefox or mozilla. If you have none of those, it will default to installing standalone nss. There is a risk that GRP binary builds will not work due to this. If it doesn't I won't hesitate to revert. Change made to net-im/gaim-1.0.3-r2 and x11-plugins/gaim-encryption-2.32-r1
Alright we've already got problems. More than likely I will revert to DEPEND on >=dev-libs/nss-3.9.2-r2
*** Bug 73035 has been marked as a duplicate of this bug. ***
Alright it turns out that Mozilla Firefox doesn't place a pkg-config file like Mozilla does. Also turns out that gaim-encryption only searches for pkg-config if the configure command-line options are not specified. However this is all moot by the fact that Mozilla isn't even keyworded for all the ARCHes that gaim is. I could ask those ARCH maintainers to either use.mask mozilla for their profiles or test/keyword net-www/mozilla. However this is an immediate problem and I am going to revert for now.
FYI that my changes have been committed to existing net-im/gaim-1.0.3-r2 and x11-plugins/gaim-encryption-2.32-r1
From what I can tell this bug was resolved just not marked that way. If I'm wrong please feel free to re-open.
