As I hate repeating myself, I'll simply refer you to bug 449550 comment 4 (and 12). With the bump, that's just become an actual problem.
*** This bug has been marked as a duplicate of bug 449550 ***
(In reply to comment #1) > > *** This bug has been marked as a duplicate of bug 449550 *** I don't think so... another issue should be tracked by another bug...
Anyway, as my real life duties are almost done, I'm now ready to check this deeper. Quoting the problems mentioned: > Seeing as the maintainer is listening, I'd like to note a few things: > - for ibus 1.5.1, the commit that came just after release > (https://github.com/ibus/ibus/commit/ > 8ac534fc002356b93e2015a8866f1ea89e3895f9) is somewhat essential > > - perhaps it's just a local problem, but "Restart" from panel menu is plain > broken: it results in input stuck to a single window (the one that was > active when the function was selected) > > - ibus-anthy compatible with >=ibus-1.4.99 will need gnome2_icon_savelist > and gnome2_icon_cache_update in correct places > > - that version also has an option for using introspection; something like > '$(use_enable deprecated pygtk2-anthy)' sounds fine > Though on the plus side, in the meanwhile this commit has been made: > https://github.com/ibus/ibus/commit/62cd0492e3459416e1775aedc327bced53f66828. > > It's hard to tell whether or not it will fix the problem with the restart - > the preliminary tests suggest not, still, it might be useful anyway. > > Note however, that this commit touches *only* gtk2 client - while there are > significant changes between gtk2 and gtk3, I don't recall any that would be > relevant here (other than multidevice support, but that's a different > matter), so it's likely that gtk3 needs this commit too (well, till this > commit the code was the same in both branches). gtk3 is also ok, because gtk3's side is actually a symlink to gtk2 one. I feel the bug subject is too obscure. We need some more specific one.
(In reply to comment #3) > Anyway, as my real life duties are almost done, I'm now ready to check this > deeper. > > Quoting the problems mentioned: > > > Note however, that this commit touches *only* gtk2 client - while there are > > significant changes between gtk2 and gtk3, I don't recall any that would be > > relevant here (other than multidevice support, but that's a different > > matter), so it's likely that gtk3 needs this commit too (well, till this > > commit the code was the same in both branches). > > gtk3 is also ok, because gtk3's side is actually a symlink to gtk2 one. > I don't follow you there - if your reply was regarding this part, that's just can't be true, as a program can't work if it's linked to both gtk2 *and* gtk3, as there's a hardcoded abort in gtk3 if mixed linking is detected. In the upstream repo the sources aren't symlinked either.
(In reply to comment #4) > (In reply to comment #3) > > gtk3 is also ok, because gtk3's side is actually a symlink to gtk2 one. > > > > I don't follow you there - if your reply was regarding this part, that's > just can't be true, as a program can't work if it's linked to both gtk2 > *and* gtk3, as there's a hardcoded abort in gtk3 if mixed linking is > detected. In the upstream repo the sources aren't symlinked either. The commit you mentioned is this: https://github.com/ibus/ibus/commit/62cd0492e3459416e1775aedc327bced53f66828 and it changes client/gtk2/ibusimcontext.c, and client/gtk3/ibusimcontext.c is seemed to be a simlink to that file. or am I missed something?? https://github.com/ibus/ibus/blob/master/client/gtk3/ibusimcontext.c
(In reply to comment #5) > (In reply to comment #4) > > (In reply to comment #3) > > > gtk3 is also ok, because gtk3's side is actually a symlink to gtk2 one. > > > > > > > The commit you mentioned is this: > https://github.com/ibus/ibus/commit/62cd0492e3459416e1775aedc327bced53f66828 > > and it changes client/gtk2/ibusimcontext.c, and client/gtk3/ibusimcontext.c > is seemed to be a simlink to that file. or am I missed something?? > > https://github.com/ibus/ibus/blob/master/client/gtk3/ibusimcontext.c OK, got confused by the tarball and github - 'git clone' indeed shows it's a symlink.
(In reply to comment #3) > Anyway, as my real life duties are almost done, I'm now ready to check this > deeper. > > Quoting the problems mentioned: > > > Seeing as the maintainer is listening, I'd like to note a few things: > > - for ibus 1.5.1, the commit that came just after release > > (https://github.com/ibus/ibus/commit/ > > 8ac534fc002356b93e2015a8866f1ea89e3895f9) is somewhat essential commit to ibus-1.5.1-r1 > > - perhaps it's just a local problem, but "Restart" from panel menu is plain > > broken: it results in input stuck to a single window (the one that was > > active when the function was selected) Cannot reproduce this here. with ibus-anthy? > > - ibus-anthy compatible with >=ibus-1.4.99 will need gnome2_icon_savelist > > and gnome2_icon_cache_update in correct places commit to ibus-anthy-1.5.0 > > - that version also has an option for using introspection; something like > > '$(use_enable deprecated pygtk2-anthy)' sounds fine ibus-anthy-1.5.0 won't work without introspection. Its python script is depending on introspection code. > > Though on the plus side, in the meanwhile this commit has been made: > > https://github.com/ibus/ibus/commit/62cd0492e3459416e1775aedc327bced53f66828. commit to ibus-1.5.1-r1
(In reply to comment #7) > (In reply to comment #3) > > Anyway, as my real life duties are almost done, I'm now ready to check this > > deeper. > > > > Quoting the problems mentioned: > > > > > - perhaps it's just a local problem, but "Restart" from panel menu is plain > > > broken: it results in input stuck to a single window (the one that was > > > active when the function was selected) > > Cannot reproduce this here. with ibus-anthy? > Yes, with ibus-anthy and a vte-based terminal (if the later matters). > > > - that version also has an option for using introspection; something like > > > '$(use_enable deprecated pygtk2-anthy)' sounds fine > > ibus-anthy-1.5.0 won't work without introspection. Its python script is > depending on introspection code. > Well, yes and no. The answer to this lies in bug 449550 comment 5. See the upstream issue I've created - that it's missing in 1.5.0 tarball was a packaging bug.
(In reply to comment #8) > (In reply to comment #7) > > (In reply to comment #3) > > > Anyway, as my real life duties are almost done, I'm now ready to check this > > > deeper. > > > > > > Quoting the problems mentioned: > > > > > > > - perhaps it's just a local problem, but "Restart" from panel menu is plain > > > > broken: it results in input stuck to a single window (the one that was > > > > active when the function was selected) > > > > Cannot reproduce this here. with ibus-anthy? > > > Yes, with ibus-anthy and a vte-based terminal (if the later matters). > > > > > - that version also has an option for using introspection; something like > > > > '$(use_enable deprecated pygtk2-anthy)' sounds fine > > > > ibus-anthy-1.5.0 won't work without introspection. Its python script is > > depending on introspection code. > > > Well, yes and no. > The answer to this lies in bug 449550 comment 5. > See the upstream issue I've created - that it's missing in 1.5.0 tarball was > a packaging bug. I've solved that with retrieving anthy.i from github and it's working fine. What I was noting was "introspection" code, not pygtk2 part. It failed to start without introspection because of the python script is depending on introspection code. Take a look at this line: https://github.com/ibus/ibus-anthy/blob/master/engine/engine.py#L46
(In reply to comment #9) > (In reply to comment #8) > > (In reply to comment #7) > > > (In reply to comment #3) > > > > > - that version also has an option for using introspection; something like > > > > > '$(use_enable deprecated pygtk2-anthy)' sounds fine > > > > > > ibus-anthy-1.5.0 won't work without introspection. Its python script is > > > depending on introspection code. > > > > > Well, yes and no. > > The answer to this lies in bug 449550 comment 5. > > See the upstream issue I've created - that it's missing in 1.5.0 tarball was > > a packaging bug. > > What I was noting was "introspection" code, not pygtk2 part. It failed to > start without introspection because of the python script is depending on > introspection code. > Take a look at this line: > https://github.com/ibus/ibus-anthy/blob/master/engine/engine.py#L46 OK, if you put it this way, it does look like a thinko on upstream part, cause it doesn't look like anything would be using that swig wrapper. Perhaps ibus-anthy 1.5.0 was initially meant to support either introspection or pygtk, but ended up being introspection only ? As I'm using ibus[-deprecated] anyway, it doesn't mater much for me. I'm more interested in that restart problem, that you couldn't reproduce. Cause while it's a rarely used option, it effectively induces a hang.
https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=40cd695c768416fadd1dba30e1e7990c28a7caa3 Drop deprecated USE flag.