[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Denemo-devel] Denemo crashes
|
From: |
Richard Shann |
|
Subject: |
Re: [Denemo-devel] Denemo crashes |
|
Date: |
Mon, 13 May 2013 17:52:54 +0100 |
On Mon, 2013-05-13 at 17:50 +0200, Éloi Rivard wrote:
> I am working with -Wall -Wno-deprecated-declarations, I will see later
> for gtk deprecation warnings.
> I am currently working on warning suppression. There is a lot of
> unused functions
There are just one or two cases I know of where functions are there for
use from a debugger, or when inserting debug statements - well, only one
that I can name immediately - it is a show_type() or some such, which
gets the name of a gtk type. Such things should be (perhaps are) always
marked as being for debugging.
> and variables, I can suppress some things, but I just comment some
> others, because I don't know if you plan to use theme someday, in this
> case I set a tag UNUSED to allow you to find it. I can also see a lot
> of #ifdef 0. Maybe some cleanup would be welcome, for readability ?
On balance I think the clean up would be good - it will be enough if I
look over the things you delete and copy out anything that looks useful,
perhaps restoring it with some more self-explanatory #ifdef rather than
the #if 0, indeed I think there are a few bits of #ifdef there that
could get activated one day - a bug in the counting of columns in
GtkTextBuffer which may be fixed in gtk3 for example.
>
> A lot of warning concern typecasting indeed, for data storage for
> example. There is some very bad code about type length (int vs char in
> 64 archs for instance), but I will solve it in a second time since it
> is less trivial.
>
> And there some cases like the original one of this email, implicit
> function declarations. Maybe that will prevent some other crashes like
> yours.
Sounds good!
Richard
>
>
>
> 2013/5/13 Richard Shann <address@hidden>
> On Mon, 2013-05-13 at 09:08 +0200, Éloi Rivard wrote:
> > That makes sense. I didn't event understand why it compiled
> at all :)
>
>
> It is very common for it not to matter, and not uncommon for
> it to fail
> consistently, but I am surprised at seeing apparently random
> crashes -
> on one run the label of one menu item was a string that
> belonged
> somewhere else altogether. I still haven't seen any crashes
> since, but I
> haven't done any serious music typesetting.
>
> In this case, if you don't give the prototype for a function
> it assumes
> the function returns an int. This only matters if it actually
> returns
> something larger - I am guessing that a pointer on a 64-bit
> architecture
> might have its top bits corrupted by the optimizer squeezing
> something
> else into what it thinks are unused bits in a register.
> >
> > I can work on warning suppression, that would maybe make the
> code
> > safer.
>
>
> There are one or two cases which I didn't get to the bottom of
> - type
> conflicts with built-in functions. And a bunch of warnings
> that arise
> from code inside macros that I couldn't quickly fix.
> And then, most worryingly there are complaints now about
> storing
> pointers in integers which I haven't looked closely into, but
> I imagine
> could be tricky to fix.
>
> >
> >
> >
> > There is a lot of warnings about gtk.
>
>
> Are you referring to the default warning level or -Wall?
>
> Richard
>
>
> >
> >
> >
> > 2013/5/12 Richard Shann <address@hidden>
> > Eloi,
> >
> > I was getting a crash when doing File->New. The
> crash vanished
> > when I
> > turned the optimizer off. I re-compiled and noticed
> that
> > keymapio.c had
> > some warnings about prototypes so I put the headers
> in
> > keymapio.h, now
> > the crash has gone away. Before finding this
> consistent crash
> > I was just
> > getting occasional crashes on startup, never the
> same twice.
> > Denemo used to be full of warnings on compile, but I
> have been
> > reducing
> > them whenever I work on a file if I can; I hope this
> is
> > another case
> > where it has paid off and that we don't have some
> threading
> > bug lurking
> > in the code.
> > Below is the backtrace FWIW.
> >
> > Richard
> >
> > #2 0x00000000004b842b in add_ui (
> > menupath=0xffffffffd802a010 <Address
> 0xffffffffd802a010
> > out of
> > bounds>,
> > after=0x0, name=0x0)
> at ../../denemo/src/keyboard.c:195
> > #3 0x00000000004b886c in create_command
> (is_script=<value
> > optimized
> > out>,
> > name=0x7fffd802a0b0 "ScoreTitle", label=<value
> optimized
> > out>,
> > scheme=0x7fffd8033ae0 ";;;ScoreTitle\n(let ((tag
> > \"ScoreTitle\")
> > (title ScoreTitle::params))\n\t(define
> (blank-title)\n\t\t
> > \t(if (not
> > (d-DirectiveGet-header-postfix tag))\n\t\t\t\t(begin
> \n\t\t\t
> > \t
> > \t(d-DirectivePut-header-override tag 0) \n\t\t\t
> \t"...,
> > tooltip=<value
> > optimized out>, hidden=0, after=0x0,
> > menupath=0x0,
> > fallback=0xffffffffd802a010 <Address
> 0xffffffffd802a010
> > out of
> > bounds>,
> > menupaths=0x0, merge=1)
> at ../../denemo/src/keyboard.c:279
> > #4 0x00000000004d9a2f in parseScripts
> (filename=<value
> > optimized out>,
> > interactive=<value optimized out>)
> > at ../../denemo/src/keymapio.c:56
> > #5 parseCommands (filename=<value optimized out>,
> > interactive=<value optimized out>)
> > at ../../denemo/src/keymapio.c:213
> > #6 parseKeymap (filename=<value optimized out>,
> > interactive=<value optimized out>)
> > at ../../denemo/src/keymapio.c:230
> > #7 load_xml_keymap (filename=<value optimized out>,
> > interactive=<value optimized out>)
> > at ../../denemo/src/keymapio.c:294
> > #8 0x000000000046cd86 in instantiate_script
> (action=0xfd5960)
> > ---Type <return> to continue, or q <return> to
> quit---
> > at ../../denemo/src/view.c:8003
> > #9 0x000000000046d0e5 in activate_script
> (action=0xfd5960,
> > param=<value optimized out>)
> > at ../../denemo/src/view.c:8082
> > #10 0x000000000046d248 in scheme_script_callback (
> > script=<value optimized out>,
> params=0x7fffdd149d80)
> > at ../../denemo/src/view.c:1080
> >
> >
> >
> > _______________________________________________
> > Denemo-devel mailing list
> > address@hidden
> > https://lists.gnu.org/mailman/listinfo/denemo-devel
> >
> >
> >
> > --
> > Éloi Rivard - address@hidden
> >
> > « On perd plus à être indécis qu'à se tromper. »
> >
>
>
>
>
>
>
> --
> Éloi Rivard - address@hidden
>
> « On perd plus à être indécis qu'à se tromper. »
>