/srv/irclogs.ubuntu.com/2007/12/03/#ubuntu-mobile.txt

=== dantalizing_ is now known as dantalizing
=== doko_ is now known as doko
dholbachgood morning08:05
loolasac: So I'm looking at dropping the nss/nspr deps when moz's version is >= 1.8 in tinymail; checking for existing usage, it seems nss / nspr is used in "libtinymail-camel" conditionally; it looks like the API is used directly10:40
loolasac: If HAVE_NSS is set, some structures get PRFileDesc members or things like this10:41
loolasac: I don't quite see how this should be pulled over xpcom; do you think this is improper use?10:41
loolasac: Hmm nevermind, I see nspr.h is in /usr/include/xulrunner-1.9b1/unstable/nspr.h10:42
loolasac: Still not sure this should be enabled or not10:43
loolasac: Argh, and MOZILLA_HOME isn't found; I don't see it in libxul-embedding-1.9.pc (tinymail uses libdir, but I see no alternative to that in the .pc either)10:54
asaclool: we will eventually migrate to use system nspr/nss11:07
loolasac: Ok, it's quite easy with the AC_ARG_WITH() flags I added11:08
asaclool: you should test for either mozilla-nspr-1.9 or nspr11:08
asac-1.9 will be dropped soon too11:09
asac(the versioning)11:09
asaclool: how do they use MOZILLA_HOME?11:09
loolasac: for gtk_moz_embed_set_path()11:12
loolWhich seems related to libxul-embedding to me11:13
loolasac: It seems to build without nspr/nss .pc files (unless it's dragged by libxul-embedding-1.9's requires or something)11:13
asaclool: gtk_moz_embed_set_path is done in glue initialization code now ... so no need for MOZILLA_HOME for you11:14
loolasac: Cool, thanks11:17
loolwidgets/.libs/libmodest-widgets.a(modest-mozembed-msg-view.o): In function `set_message':11:19
lool/home/lool/scratch/modest/modest-1.0/src/widgets/modest-mozembed-msg-view.cpp:673: undefined reference to `modest_tny_msg_find_body_part(_TnyMsg*, int)'11:19
loolCool, I think I get the errors StevenK was getting now11:19
asac:)11:19
StevenKThat'd be them11:21
StevenKAny help gratefully received11:21
loolStevenK: It's a circular dep between src/ and src/widgets I'm afraid11:24
asaclool: so is the nss/nspr issue mentioned in mail now resolved?11:25
loolsrc/widgets/modest-mozembed-msg-view.cpp uses modest_tny_msg_find_body_part() which is in src/modest-tny-msg.c and src/Makefile.am asks to link modest against widgets/libmodest-widgets.la11:25
loolasac: Well theoritically I don't use nss/nspr modules as I pass --with-mozilla-nss-pc=no, but technically there are many refs to nss/nspr and I didn't review them al11:26
loolall11:26
loolasac: It looks like some extensive work to properly review and cleanup the whole tree to me  :-/11:27
loolEh modest is Copyright (c) 2006, Nokia Corporation11:28
asaclool: you can use nspr11:29
asacand nss11:29
loolasac: You think I should if it builds withoutN11:29
lool?11:29
asacright ... its better to not do that, but if its needed it should be possible :)11:30
loolasac: Sure; for now it seems it builds without, it's trivial to add the b-deps and pass the proper --with if necessary11:30
loolStevenK: dpatch updated for tinymail BTW (pub/~lool)11:30
loolerr p.u.c/~lool11:30
loolStevenK: So looking at it, it's no fun: 32 widgets sources, not an option to move, src/modest-tny-msg.c depends on -folder and -text-utils which probably depend on other stuff => might be movable, and modest itself is clearly not movable11:32
MithrandirStevenK: about helix, do you think there is any development work we need to do about it, or is it just getting the integration working?11:33
MithrandirStevenK: I'm trying to make up my mind wrt minispecs/a full spec for it.11:33
StevenKlool: Don't bother with moving, I just want it to compile.11:37
StevenKMithrandir: Um. If I could remember the mail that said what needed to be done for helix11:38
loolStevenK: Moving is usually the long term way to fix circular deps across automake dirs AFAIK11:38
loolBlah SVN modest needs SVN libwp11:40
StevenKRight11:40
=== dholbach_ is now known as dholbach
StevenKlool: I saw your mail about tinymail, let's see what Philip says11:41
loolStevenK: Can you confirm where the libwpeditor SVN is?11:42
loolATM I'm guessing in garage along modest11:42
StevenKhttps://garage.maemo.org/svn/modest/libwpeditor-plus/trunk11:42
loolOk, that's what I found as well, thanks11:43
loolwp_text_buffer_replace_image isn't even in libwp SVN, sigh11:45
loolAnd source is missing at repository.maemo.org, cool11:48
StevenKRight. I commented that bit out11:50
loolI'll write the Nokia guy who wrote the change that he didn't update libwp11:54
agoliveiralool, StevenK, welcome to the wonderfull world of maemo repos hell :)11:56
StevenKHaha11:56
loolagoliveira: Hehe11:59
asaclool: i think mozilla_xpcom_includetype needs to be empty when $mozilla_xpcom version is < 1.911:59
asaclool: what about the GLUE patch?12:00
loolasac: That one I wanted to think more about it and rework, it's a bit ugly ATM12:01
loolasac: You mean the one in configure.ac, right?12:01
loolasac: I'd like the flags to be pulled cleanly   :-/12:01
asacno the one in code ... mozembed_init 12:02
asacor something like that12:02
asacbut the one in configure.ac needs to be fixed as well12:02
asac#ifdef XPCOM_GLUE should be clean enough imo12:02
loolasac: Oh yeah, also a separate fix, but didn't send it yet indeed12:04
asacah ok12:04
loolasac: You want to send it?12:04
loolI didn't have to change it in anyway since your version12:04
loolhaha the Nokia guy is running gutsy :)12:06
loolHe commits his debian/changelog with +modest (1.0-maemo21) gutsy; urgency=low12:06
asacmakes sense to use gutsy ;)12:06
MithrandirJan-Dirk or somebody else?12:06
loolYeah, this guy12:07
StevenKRight12:07
loolAnother mail sent for the second error12:09
loolmodest-mozembed-msg-view.c:49:40: error: widgets/modest-scroll-area.h: No such file or directory12:09
loolAnd then modest-mozembed-msg-view.c:373: warning: implicit declaration of function 'modest_scroll_area_new'12:09
asaclool: imo the trunk branch is broken12:09
looland obvious failure after that12:09
asaci ran into the same issues12:09
asacbut could build with gnomeui12:09
loolasac: You had to build tinymail with gnome-desktop stuff though, right?12:09
loolIt really looks like modest is in constant flux and quite rough in SVN12:10
asachmm ... no idea ... the gnomeui was just a joke :) ... no functionality, mostly dialogs saying "Hello" opened when activating some menu/button12:10
loolasac: ah :-(12:10
asacwhy do you want trunk=12:11
asac?12:11
loolIn file included from ../../src/widgets/modest-mozembed-mime-part-view.h:40, from modest-mozembed-msg-view.c:48:12:12
lool/usr/include/libtinymailui-mozembed-1.0/tny-moz-embed-html-mime-part-view.h:28:25: error: gtkmozembed.h: No such file or directory12:12
loolasac: Why search for bugs which might be fixed in trunk or for which the fixes might not apply to trunk?12:12
loolI usually try reproducing stuff with trunk/head, if I do I fix it there, then backport downstream12:12
loolPff the -I flags when building tinymail take two xterm screens, it's ridiculous12:13
StevenKI think half of our problems with modest are because of xul embedding requiring that evil evil hack of .c -> .cpp12:13
loolasac: Ah, I have an issue for you here12:14
loolMODEST_MOZEMBED_CFLAGS ends up with -I/usr/include/xulrunner-1.9b1/stable12:14
loolIt should be unstable12:14
loolSo I guess libtinymail should be setting includetype?12:15
loolThe thing is that pkg-config var are in a common namespace, so it's a bit ugly to set includetype in tinymail's .pc12:15
asacyes 12:15
asache?12:16
loolWould it make sense to rename includetype to libxul-embedding-includetype?12:16
asaclool: that would need to be done upstream12:16
loolasac: That's what I was suggesting :)12:16
loolasac: Unlikely to ever happen?12:17
asacwell ... i will ask :)12:17
loolIt's a concrete example: a lib exposing it's gtkmozembed implementation needs to set this var12:17
loolasac: Cool, thanks :)12:17
asacbut he isn't awake yet ... so later12:17
loolI need to get some food before 2pm &12:18
Mithrandiragoliveira: I'm pondering renaming mobile-applications to mobile-core-applications and make mobile-applications the full set of apps from the SoW, using the minispec idea.  Thoughts on that?12:24
agoliveiraMithrandir: I guess it's ok.12:25
agoliveiraMithrandir: and actually makes sense if we want to consider the whole lot of applications.12:25
loolMithrandir: Do you know whether a .pc can define vars to be used in another one?13:32
Mithrandirlool: I don't believe that is supported, no.13:32
loolMithrandir: Like libtinymailui-mozembed-1.0 would set includetype=unstable so that the require would use it13:32
loolMithrandir: Ok, thanks13:32
Mithrandirship two .pc files.13:32
loolMithrandir: Exactly what I was about to write to asac  :)13:32
loolasac: ^13:32
loolasac: Would be nice to have libxul-embedding-unstable and libxul-unstable13:33
loolasac: Because I can't set the var in libtinymailui-mozembed-1.0, I tried it's not honoured13:33
Mithrandirto me, it looks like the mozilla people are trying to bend pkg-config to do something it's not designed to do (and which it isn't easily extended to do)13:33
loolMithrandir: Eh, it's Mozilla!   :-P13:34
asacpkg-config can do it ... the macros can't though :) ... which doesn't make much a difference13:34
asac... i will ask if they accept a patch for -unstable :)13:34
asacoh ... misread. nm13:35
asaclool: http://pastebin.mozilla.org/255146 ... appears to use unstable 13:39
loolasac: Yes, but not if I put it in libtinymailui-mozembed-1.0.pc13:39
loolasac: But it's of course fine if I set it on the command line13:39
asacwhat do you mean? i run pkg-config on libtinymailui-mozembed-1.0 in that paste13:39
loolasac: It's just unpractical to change all PKG_CHECK_MODULES() in other packages into $PKG_CONFIG calls13:39
asacah13:39
asacok13:40
loolIt kind of defeats the pkg-config concept :-/13:40
asacwell but why is it inherited when using --define-variable ... but not if you define the variable in the .pc file itself?13:40
asacsounds like a bug as well13:40
Mithrandirif anything, it should be something like Requires: libxul-embedding-1.9 [includetype=unstable] in the .pc file13:40
Mithrandirasac: no, that's not a bug.  There's nothing which implies that a .pc has a way of reaching out of its own scope.13:41
asacyes right. bad thought :)13:43
loolOk, passed one more failure in modest SVN, now at modest-mozembed-msg-view.c:385: warning: implicit declaration of function 'gtk_widget_tap_and_hold_setup'13:57
asaclool: its a #define gtk_widget_tap_and_hold_setup(a,b,c,d)14:00
asacand might be excluded because of #ifndef MODEST_HAVE_HILDON_GTK14:00
* asac out for lunch14:01
Mithrandirlool: that one we need to patch around, since we don't have tap-and-hold yet.14:01
loolMithrandir: Yup, i'm adding an ac_check_decl in modest14:03
loolasac: good lunch14:03
loolasac: I think I prefer changing their #ifdef maemo_chages which is set for some reason to a more specific check14:04
loolCool, fails to link now14:14
StevenKDamn modest14:15
loolHmm if a C file calls gtk_moz_embed_find_text(), I do need to link to -lxul, no?14:20
=== cprov is now known as cprov-lunch
asaclool: no14:38
asaclool: well ... i can't tell for sure, epiphany, devhelp, yelp ... all use typeaheadfind not the mozembed function14:41
asacbut i would guess that its provided by gtkmozembed_glue.cpp14:42
loolsmagoun: Heya, around?15:16
loolsmagoun: I read about your freetype1 changes, but couldn't see the difference between your package in the ppa and the one in hardy (apart of the build-conflicts); is this fixed in hardy?15:16
smagounlool: hi15:16
smagounlool: I think it's fixed in hardy, yup.15:17
StevenKsmagoun: You told me it was15:17
smagounI didn't see any sign of a fix for gutsy, even though it was clearly broken - hence the upload to the PPA15:17
smagounStevenK: The bug (let me find it...) mentioned a Hardy fix15:17
loolsmagoun: Ok, everything is fine then15:19
loolsmagoun: I just wanted to make sure that it was fixed in hardy, I couldn't see any difference between hardy and ppa build-deps, so it's fine15:19
smagounHere's the bug: https://bugs.edge.launchpad.net/ubuntu/+source/freetype1/+bug/13566315:20
ubotuLaunchpad bug 135663 in acton "FTBFS: wrong build-dependency on libkpathsea4-dev" [Undecided,Fix committed] 15:20
StevenKAh yes, I remember that one15:20
StevenKAnyway, I'm going to bed, since it's 2:2015:20
smagounStevenK: 'night15:20
StevenKNight15:20
smagounIs there a way to see all packages that failed to build for gutsy/LPIA? We've run into this problem a couple times now.15:21
smagounIt'd be nice to know what else is broken15:21
StevenKlaunchpad.net/ubuntu/gutsy/lpia/+builds15:21
smagoun485 build failures. Hmm.15:22
StevenKEnjoy :-P15:22
smagounPresumably some were superseded by working builds?15:22
StevenKRight15:22
smagounugh15:22
StevenKHence the enjoy. :-P15:22
StevenKAnyway, night 15:22
loolsmagoun: Better to fix them when you need to  ;)15:23
loolsmagoun: Ok, thanks for the bug id15:23
smagounlool: The problem is that we don't even know what's broken/out-of-date!15:23
smagoungrep -v kde isn't a valid input to the package filter on that page....bummer15:25
loolsmagoun: Ah you want to compare the lpia versions with the i386 versions?15:25
smagounlool: yeah15:25
loolsmagoun: You can script that over the Packages file easily15:25
loolgrep-dctrl provides tools to extract any field you like out of the files, but even awk should be enough I guess15:26
smagounno kidding. I'll have to take a look, I didn't know about grep-dctrl15:26
* smagoun is allergic to awk15:26
smagounThanks!15:27
loolTry perl!15:29
smagoun:)15:29
loolActually sort / uniq would be enough :)15:29
=== cprov-lunch is now known as cprov
loolasac: Do I understand correctly that it should be libtinymailui-mozembed-1.0 which should #include gtkmozembed_glue.cpp?16:32
loolHmm no that's currently fine16:34
loolasac: If I #include <gtkmozembed_glue.cpp> just before gtkmozembed.h, just like in the yelp example in the wiki, I get a /usr/include/xulrunner-1.9b1/unstable/nscore.h:117:1: error: "NS_HIDDEN" redefined16:38
loolOh it's actually a problem with gtkmozembed_glue.cpp alone, the two includes are from gtkmozembed_glue.cpp16:41
asaclool: you have to compile with g++ ... are you doing that?16:52
asaclool: otherwise you have wrong CFLAGS16:52
loolasac: Ah no16:52
loolasac: Is there a solution to build with gcc?16:52
asac(repeated: in the end the zooming should encapsulated in libtinymailui-mozembed)16:53
loolasac: Or do you have pointers to build with g++?16:53
asaclool: just rename the file extension of the .c file that now includes the .cpp file to 16:53
loolasac: I don't understand your last sentence16:53
asac.cpp16:53
lool"repeated: ..." => what do you mean?16:53
asaclool: libtinymailui-mozembed implements the whole gecko business :)16:54
asacnow modest people hacked because they wanted zoom 16:54
loolOh you mean modest should do gecko directly?16:54
asacno ... it shouldn't ... but it currently does :)16:54
loolOk16:54
loolasac: Hmm I renamed the .c file including tny-moz-embed to .cpp, finally got g++ to build it, but it fails in the same way17:16
loolAh, it's not the same file I think, sorry17:16
loolasac: I'm giving up, can I hand you the ball?17:33
asacwhats happening now?17:34
loolSame thing17:34
loolExcept I'm building with g++17:34
asacis -DXPCOM_GLUE defined?17:34
loolIt is17:34
asaclool: did you include the _glue.cpp in two files?17:35
loolI'll copy the last lines of build to you17:35
loolasac: I did17:35
asaclool: that won't work17:35
asaclool: just once17:35
asacit just exists to make the symbols available during linking17:35
loolhttp://paste.debian.net/4407917:36
loolasac: Ah17:36
asacgtkmozembed.h is enough for just the headers17:36
loolasac: Looks better; *sigh* thanks a lot17:37
loolasac: How shoud I pick the file where to include the cpp?17:37
asaclool: usually the right place is to do it wherever the previous code intiialized gecko17:38
loolasac: And am I supposed to rename all files including mozembed to cpp?17:38
asace.g. gtk_moz_embed_set_path17:38
loolShould I protect with ifdef XPCOM_GLUE?17:39
asacyes17:39
loolAm I supposed to rename all files including mozembed.h to .cpp?17:39
asacno ... I think just the file with the _glue.cpp17:40
asaclool: anyway, if you want to hand it over, just open a xulrunner 1.9 bug to track this. and tell me the svn branch, the build-depends and the configure flags of modest and tinymail + whatever patches you already have :)17:41
loolOk, I will17:42
loolasac: If I don't rename the other file, I get the error17:42
loolAh no wait, automake needs a distclean again grrr17:42
loolautotools have been very painful when switching a file with the same name from C to CPP17:43
asacyes :) ... run sh autogen.sh17:43
loolIt's not enough, I have to cleanup some dependency files manually17:43
lool(To properly cleanup, I'd have to distclean before renaming)17:43
asacoh right .. if you forgot to clean before you run into issues17:44
asaclool: if you ask for help on xul 1.9 could you at least try if the modest trunk really works fine with 1.8?17:45
loolasac: Ok17:45
asaci just fear to work against something that doesn't work at all17:46
loolasac: http://paste.debian.net/4408017:47
loolXPCOM_GLUE is defined; this only cpp fiel is where the init is supposed to happen, and it produces this despite the mozembed.cpp include17:47
asaclool: isn't that just a warning?17:48
loolasac: It is, shall I ignore that?17:48
asacyes17:48
asacthats fine17:48
loolWow, I've been fighting such warnings most of this afternoon thinking these were the issue17:49
asac(well fine is something else, but i never had issues because of that)17:49
asacoh sorry ... i thought the paste before was an error:17:49
loolasac: It was, but I also got warnings oh well17:50
loolWouah SVN built after the last error I fixed before dinner19:59
=== cprov is now known as cprov-out
=== asac_ is now known as asac
loolStevenK: So, SVN builds for me now \o/20:32
loolStevenK: The circular dep must have been temporary and fixed upstream (would not be surprizing); I'm sending my patches upstream and I shall prepare an updated snapshot tomorrow, with the fixes20:33
loolStevenK: How could I easily extract your changes to reapply them to the snapshot as to not lose any of them?20:33
loolasac: Thanks for your help BTW, I think I was losing patience at the end of the day with the many Mozilla issues which are really me learning how to use their stuff properly20:35
loolWow, serious bug we have since a long while: Variable 'datarootdir' not defined in '/usr/lib/pkgconfig/osso-af-settings.pc'20:56
loolCool, saw my first modest window before it crashed \o/20:59
=== sabotage is now known as sabotage_afk

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!