Help save net neutrality! Learn more.
Close

CVS' "Toolkit mismatch" but not with 2.60

Help
ksandre
2004-05-27
2004-06-03
  • ksandre

    ksandre - 2004-05-27

    When attempting to view a clip using the CVS version of the mplayerplug-in [05/26/04], I get the following:

    "Toolkit mismatch mozilla(GTK3080302), plugin(GTK1)"

    with no successful play of the clip.

    Previously I have successfully used the version 2.60 without a problem (except by my own inept limitations).

    "ldd" on mozilla-bin does, indeed, show that Gtk1 is enabled (not Gtk2).

    First built the CVS with "--with-gecko-sdk=<path>" and "--enable-gtk1" then after first error I tried adding "--disable-gkt2" with the same result.

    I have not yet tried to use "--with-x" instead of "--enable-gtk1" because I thought I recalled seeing similiar messages here about this error, but I can not find same just now.

    I am using an older version of Mozilla [mozilla-i686-pc-linux-gnu-egcs112-1.4] with the gecko-sdk-i686-pc-linux-gnu-1.6 [Mozilla1.6] which was reported to work with older Mozilla versions at the mplayerplug-in Web site.  As I said, I have been successfully viewing with this arrangement for awhile now.

    Maybe I have just been lucky in avoiding error messages all this time, eh?

    Additionally, I have also just now changed from MPlayer-1.0pre4 to the CVS snapshot [05/26/04].
    (I also have both Gtk1 and Gtk2 installed on my system.  Current gcc is 2.95.3 and glib 2.2.5.)

    Thank you!  I have been using this plugin for a LONG time now [through many versions] and have had very good luck.  Keep up the excellent work!

    =====

    extra information from my build logs:
    ---------------------------------------

    Source/plugin.cpp: In method `NPError nsPluginInstance::SetWindow(NPWindow *)':
    Source/plugin.cpp:486: warning: unsigned int format, Window arg (arg 2)
    Source/plugin.cpp:486: warning: unsigned int format, pointer arg (arg 3)
    Source/plugin.cpp:511: warning: unsigned int format, pointer arg (arg 4)

    Source/plugin-list.cpp: In function `void deleteList(Node *)':
    Source/plugin-list.cpp:176: warning: int format, long int arg (arg 2)
    Source/plugin-list.cpp: In function `void printNode(Node *)':
    Source/plugin-list.cpp:808: warning: int format, long int arg (arg 2)

    Source/plugin-threads.cpp: In function `int playNode(ThreadData *, Node *, char *, int, int *, int *
    )':
    Source/plugin-threads.cpp:664: warning: unknown conversion type character `(' in format
    Source/plugin-threads.cpp:566: warning: `char * eos' might be used uninitialized in this function

    ======

     
    • Kevin DeKorte

      Kevin DeKorte - 2004-05-27

      The toolkit mismatch warning / error was recently added to help people compile mplayerplug-in correctly. Many people have had crashes just from picking the wrong toolkit.

      I would recommend upgrading your browser to Mozilla 1.6 or Firefox 0.8 and see if that helps. It looks to me that Mozilla is reporting the wrong toolkit or a datacoded release. As the toolkit info it obtained directly from mozilla.

      Or to get around this error without changing your browser at about line 485 in plugin-support.c add a line that says "return 0;" and that will skip the tests.

      I'll look at the other warnings and see if I can clean them up.

       
    • ksandre

      ksandre - 2004-05-28

      I had erroneously said:  "I have not yet tried to use "--with-x" instead of "--enable-gtk1"...."  I meant to say "--enable-x" instead.

      So I did this, and it works now.  Athough I do notice some of the buffering/download window bleedthroughs and overcasts during playback that have been noted in others' posts.  I have not fiddled with it enough yet to give which-does-what-when details.

      The "warning[s]" in my build logs for this try were the same with the additional:

      ------------------------------------------

      Source/plugin-ui.cpp: In function `void DrawUI(_WidgetRec *, nsPluginInstance *, char *, int, int)':

      Source/plugin-ui.cpp:244: warning: int format, Widget arg (arg 2)

      ------------------------------------------

      kdekorte said:
      "I would recommend upgrading your browser to Mozilla 1.6 or Firefox 0.8 and see if that helps.

      I have been hesitant to upgrade Mozilla to a gcc3+ build [at version 1.4] because of my gcc2.95.3/glibc2.2.5 system and because at the Mozilla1.4 time period I had noticed both the Java incompatibility and the inability to view Flash clips.  I understand now that it is possible to use Java plugin, but still not Flash [?].  There has not been another "egcs" Mozilla build since 1.4 inspite of the pointers subsequent releasenotes.  (I keep forgetting to write the Mozilla folks and show them that the link in the notes leads nowhere, that the _only_ "egcs" build is in the 1.4 release directory.)  I have not seriously entertained the idea of attempting to build Mozilla myself, and, while I do build alot of applications myself I am still not up to rebuilding my whole system [Slackware8.1] as a gcc/glibc upgrade would require.  Probably within a month [expect a new release of Slackware when I will purchase the CDs] I will have upgraded gcc/glibc and will no longer have this issue (except for Flash).  These concerns extended to FireFox as well.

      kdekorte said:
      "It looks to me that Mozilla is reporting the wrong toolkit or a datacoded release. As the toolkit info it obtained directly from mozilla."

      -------------------

      about:buildconfig

      Build platform
      target
      i686-pc-linux-gnu

      Build tools
      Compiler     Version     Compiler flags
      gcc     gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)     -Wall -W -Wno-unused -Wpointer-arith -Wcast-align -pedantic -Wno-long-long -pthread -pipe
      c++     gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)     -fno-rtti -fno-exceptions -Wall -Wconversion -Wpointer-arith -Wcast-align -Woverloaded-virtual -Wsynth -Wno-ctor-dtor-privacy -pedantic -Wno-long-long -pthread -pipe -I/usr/X11R6/include

      Configure arguments
      --enable-extensions=default --enable-crypto --disable-tests --disable-debug --enable-elf-dynstr-gc --disable-logging --without-system-nspr --without-system-zlib --without-system-jpeg --without-system-png --without-system-mng '--enable-optimize=-O\ -g'

      ------------------------------

      kdekorte said:
      "...get around this error without changing your browser at about line 485 in plugin-support.c add a line that says "return 0;" and that will skip the tests...."

      Cool.  I will try that next time I rebuild.

      Thank you.

      I will see if I can glean some details about the "buffering window" overlapping and post later. 

       
    • ksandre

      ksandre - 2004-05-28

      [still using CVS-05/26/04  build with option "--enable-x"]

      FYI, the "buffering/download window bleedthroughs and overcasts during playback" I mention above is pretty well discribed in the thread cited below.

      SourceForge.net: mplayerplug-in-devel

      Thread title:  "some non working sites"
        
      Thread ID:
      -----------
      https://sourceforge.net/mailarchive/forum.php?thread_id=4336144&forum_id=21800

      Original message:
      -------------------
      https://sourceforge.net/mailarchive/message.php?msg_id=7927754

      Also, there are some maximise oddities and [non-] return to original size that I notice, though I still have not had time to play with it much yet as I just reread all the threads since release 2.60.

      I will try to be more specific later.

       
    • ksandre

      ksandre - 2004-06-02

      Re:  CVS plugin version with "--enable-gtk1"

      kdekorte said:

      "...get around this error without changing your browser at about line 485 in plugin-support.c add a line that says "return 0;" and that will skip the tests...."

      Doing this resulted in a "make" error [parse / syntax - sorry for lack of detail].  I tried a few variations, all failed.  I am not clueful enough regarding programming to know howto adequately compensate for the results of source changes, btw (ie., I would not necessarily know what else might need a chnage).

      I have reverted to the mplayerplug-in 2.60 release version for now.

      Re:  CVS mplugin version with "--enable-x"

      Regarding the descriptions in previous posts about the CVS version plugin compiled with "--enable-x" rather than "--enable-gtk1," and the "buffer/download" window covering the playback display.  This is consistant enough to make it impossible to use.  It is LEAST likely to occur when the clip is Quicktime, as I recall, but seems to depend on the Web site.  Other problems have been enumerated in other peoples' thread and I find simiiiar results.  There are certain types of pages employing "Flash" that will not play clips at all (and do not give an error).  I will start another thread (though I will look here first, as I seem to recall someone else mentioning this problem).

      BTW, I have that a good place to test is PBS' "Frontline" site because the newer clips available have all [ie., RP, WM, and QT] choices.  The only drawback is that the popup will autodefault to RP, so you have to immediately stop the clip [right click on playback window & press stop] and change to another player from the top line of the plugin.  Sometimes you might to reload after change.  (The drawback to the site is that it is BUSY.)

       
    • Kevin DeKorte

      Kevin DeKorte - 2004-06-03

      I'm putting a patch into the code that if the toolkit is not recognized it will give a warning about a possible toolkit mismatch to the console but it will keep running. But not being recognized it could result in mozilla crashing if the wrong toolkit is not being used.

       

Log in to post a comment.