Summary: | media-sound/clementine with >=dev-db/sqlite-3.8 - searching is extremely slow | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Nikos Chantziaras <realnc> |
Component: | Current packages | Assignee: | Nikoli <nikoli> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | adaptee, arfrever.fta, chain, dschridde+gentoobugs, Jens.Rutschmann, proxy-maint, sound, ssuominen |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
URL: | https://github.com/clementine-player/Clementine/issues/3847 | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 488164 |
Description
Nikos Chantziaras
2013-09-17 17:48:06 UTC
Note from Clementine upstream: "Yes - this is a debian packaging bug. They either need to switch STATIC_SQLITE=ON or ensure that the qt4 sqlite plugin is built with all the sqlite3 symbols visible. This is not something we can fix." Copying the ebuild to my local overlay and changing: -DSTATIC_SQLITE=OFF to: -DSTATIC_SQLITE=ON fixes the problem with sqlite-3.8. Now the upstream forces STATIC_SQLITE in the lastest code: if(STATIC_SQLITE) message(STATUS "Building static qsqlite plugin") add_subdirectory(3rdparty/qsqlite) include_directories("3rdparty/qsqlite") else() if (NOT I_HATE_MY_USERS) message(FATAL_ERROR "No, really, Clementine needs sqlite to be linked " "statically. If it's not, Clementine won't be able to resolve the " "functions it needs to register full text search support. Search will " "be really slow and users will be unhappy. If you're packaging " "Clementine for a distribution and you really hate your users and want " "them to have a bad time please rerun cmake with -DI_HATE_MY_USERS=ON") else() if (NOT MY_USERS_WILL_SUFFER_BECAUSE_OF_ME) message(FATAL_ERROR "So you really hate your users and you want to " "create a crippled package that doesn't work properly? Please send " "an email to the devs and help us understand why.") endif() endif() endif(STATIC_SQLITE) Asked clementine devs, would like to know opinions of qt and sqlite package maintainers about missing symbols. What does "ensure that the qt4 sqlite plugin is built with all the sqlite3 symbols visible" mean? It doesn't make much sense to me... if clementine needs some sqlite3 symbols it should explicitly link against it and use the symbols... I agree, it sounds like underlinking. Removing qt@ from CC wrt comment #5 and #6. Please re-add if there's something for us to do. Upstream moved to github. I just downgraded my system to 3.7.17 and this fixes things. Wouldn't that be a solution for this problem? Or is there something else that could be done on ebuild level? (In reply to Richard H. from comment #9) > I just downgraded my system to 3.7.17 and this fixes things. Wouldn't that > be a solution for this problem? Or is there something else that could be > done on ebuild level? This is already fixes in the ebuild. Also, sqlite 3.7.17 isn't even in portage anymore. Are you sure your portage tree is synced and up to date? (In reply to Nikos Chantziaras from comment #10) > (In reply to Richard H. from comment #9) > > I just downgraded my system to 3.7.17 and this fixes things. Wouldn't that > > be a solution for this problem? Or is there something else that could be > > done on ebuild level? > > This is already fixes in the ebuild. Also, sqlite 3.7.17 isn't even in > portage anymore. Are you sure your portage tree is synced and up to date? Yes, of course :) I pulled the 3.7.17 ebuild from CVS and put it into my overlay. Which ebuild has this fixes? Because the stable 1.1.1 sure does not have them. (In reply to Richard H. from comment #11) > Yes, of course :) I pulled the 3.7.17 ebuild from CVS and put it into my > overlay. Which ebuild has this fixes? Because the stable 1.1.1 sure does not > have them. 1.2.0, 1.2.1 and 1.2.2 do have the fix. Confirmed. 1.2.2 does not have this bug anymore (of course only if you don't set system-sqlite USE flag). So maybe the ebuild should get a bump then? Time to close the bug then, I presume (In reply to Richard H. from comment #13) > Confirmed. 1.2.2 does not have this bug anymore (of course only if you don't > set system-sqlite USE flag). So maybe the ebuild should get a bump then? If you're not on ~arch, you should file a stable request for 1.2.2. I believe you can do that as a normal user if you need a package from ~arch that fixes a bug. |