Summary: | mail-client/thunderbird fails to build with sys-libs/musl | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | tt_1 <herrtimson> |
Component: | [OLD] Unspecified | Assignee: | Gentoo musl team <musl> |
Status: | RESOLVED OBSOLETE | ||
Severity: | normal | CC: | blueness, mozilla |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 430702 | ||
Attachments: |
split patch for the overlay
checksum fails bump to 38.5.0 |
It'll be on the overlay in a minute. Let's work toward getting the fixes upstream. (In reply to tt_1 from comment #0) > Created attachment 415830 [details, diff] [details, diff] > split patch for the overlay > A recommendation about producing patches with git: 1) add the ebuild and accompanying files to the directory then do `git add .` 2) make sure everything you want was added with `git status` 3) commit using `git commit -m "mail-client/thunderbird: initial commit, bug #564586."` 4) produce a patch using `git format-patch HEAD^` Notice this produces a much nicer patch with a commit message, attribution etc. It can also be emailed using `git send-email` so its very nice. Created attachment 415832 [details, diff]
checksum fails
don't know why the checksum is failing from the overlay?
(In reply to tt_1 from comment #3) > Created attachment 415832 [details, diff] [details, diff] > checksum fails > > don't know why the checksum is failing from the overlay? i fixed it. please check. Most of the patches are taken from upstream anyway, so the problems are known and mostly fixed. I backported them from the main branch, as thunderbird uses firefox-esr. See the comments in the epatch section of the ebuild I made. However, the fix-netwerk.patch I took from Alpine lacks a proper description, so I have no idea wether upstream knows about it or not. Also I wasn't sure about what to do with the fix-bug-1130175.patch , as there are major changes in mozilla/media/mtransport/third_party/nICEr/src/stun/addrs.c , and I couldn't understand the logic of it. So can please someone review this patch and find out, if it is still needed for thunderbird? Created attachment 420976 [details, diff]
bump to 38.5.0
The patchset is now included in the in tree version, so I took the ebuild and edited it in terms of gnupg which if version <2.1 depends on dev-libs/pth and which doesn't compile on musl. Therefore, I made depend thunderbird on >=gnupg-2.1 because here it depends on dev-libs/npth which does compile fine on musl.
(In reply to tt_1 from comment #6) > Created attachment 420976 [details, diff] [details, diff] > bump to 38.5.0 > > The patchset is now included in the in tree version, so I took the ebuild > and edited it in terms of gnupg which if version <2.1 depends on > dev-libs/pth and which doesn't compile on musl. Therefore, I made depend > thunderbird on >=gnupg-2.1 because here it depends on dev-libs/npth which > does compile fine on musl. please attach a git patch, there is no revision bump needed just need a clean look at the changes you are requesting without having to read an entire ebuild. You can use anongit for this as a normal user. Is it possible to mask dev-libs/pth and all related version of app-crypt/gnupg which depends on it in the package.mask of the musl profile? Would be the easiest way to deal with this, I guess. *** Bug 577280 has been marked as a duplicate of this bug. *** |
Created attachment 415830 [details, diff] split patch for the overlay tested with USE="minimal jemalloc3 dbus" , for USE="crypt" one will have to use at least gnupg 2.1.8, as pth does not compile with musl. Instead, gnupg uses npth from 2.1.8 on, so I changed that in the ebuild accordingly. anyway, please review and test.