Summary: | app-crypt/seahorse{,-plugins} doesn't build against gpgme with lfs (large file support) | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Wilken Haase <Hibbelharry> |
Component: | [OLD] GNOME | Assignee: | Gentoo Linux Gnome Desktop Team <gnome> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | al_91dak, daveswen, fauli, ikelos, judepereira, nebojsa |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 277890 | ||
Attachments: | LFS patch |
Description
Wilken Haase
2009-06-25 23:43:36 UTC
how did you build gpgme with lfs support in the first place ? (In reply to comment #1) > how did you build gpgme with lfs support in the first place ? > I don't think i did something special to gpgme. After a little research i found evidence that lfs support for gpgme is enabled by default on supported systems, see here: http://pyme.sourceforge.net/doc/gpgme/Largefile-Support-_0028LFS_0029.html in /usr/include/gpgme/gpgme.h, line 78, i've replaced: #if 64 by #if (_FILE_OFFSET_BITS) == (64) and now compile. it's ok to do this? same error here :'( (In reply to comment #4) > same error here :'( > With GPGME-1.1.8-r1 it works fine (In reply to comment #3) > in /usr/include/gpgme/gpgme.h, line 78, i've replaced: > > #if 64 > > by > > #if (_FILE_OFFSET_BITS) == (64) > > and now compile. it's ok to do this? > Don't touch the header file of gpgme yourself, it's more complicated... To be clear and speak with simple words, seahorse compiles correctly with gpgme-1.1.8 because it doesn't match _FILE_OFFSET_BITS macro into gpgme.h, this match is only available since 1.2.0 version, and because gpgme-config and seahorse source code don't define this macro, the compilation fails (with 1.2.0). But _FILE_OFFSET_BITS needs to be defined to 64 on 32 bits system, to be sure that largefile (usually bigger that 2GB) will be handle correctly, otherwises the behaviour of ftell,fseek (or something else of the same family) can be ramdom or in the worst case disastrous. I'll discuss about that with my herd (gnome). However if another dev has a better idea than mine, feel free to commented out this bug. hey, i have the same error, i did what he said to change line 78, compiled it successfully, then re changed it back to the original... so now is this fine or .... Modifying the headers might or might not work; seahorse is being build with LFS, just the include orders were wrong, so that the gpgme.h didn't know it. I've fixed it in portage without a bump. If you modified the headers, you might want to re-emerge when this hits the tree. I reopen, because the current stable is broken, too and gpgme-1.2.0 is about to go stable. Can you fix the 2.24 series, please? (In reply to comment #9) > I reopen, because the current stable is broken, too and gpgme-1.2.0 is about to > go stable. Can you fix the 2.24 series, please? > ...and seahorse-plugins as well, please? Stabilization of app-crypt/seahorse-2.26.2 will occur in bug #281427. why closing this bug while user actually reported something was missing to mark it fixed... seahorse and seahorse plugins are affected, 2.24 versions fix is not mandatory since it's masked iirc. *** Bug 281883 has been marked as a duplicate of this bug. *** Comment #8 suggested that this bug is fixed. (In reply to comment #15) > Comment #8 suggested that this bug is fixed. > It is for seahorse. It is NOT fixed for seahorse-plugins Created attachment 202016 [details, diff] LFS patch This patch seems to fix compile issues with LFS in gpgme. The original patch can be found here : http://sisyphus.ru/br/srpm/Sisyphus/seahorse-plugins/patches/0 Fixed without a bump. Thanks for reporting. Ebuilds are missing "inherit eutils". Please reopen. Just looked via cvs (thanks to scarabeus for pointing that out), it's all fine now. reopen per comment 19 +04 Sep 2009; Romain Perier <mrpouet@gentoo.org> + seahorse-plugins-2.26.2.ebuild: + Fix missing inheritance from eutils, per bug #275445. Fixed into the main tree without a bump. *** Bug 287669 has been marked as a duplicate of this bug. *** |