Re: Crash removing Format

2021-03-09 Thread Scott Kostyshak
On Tue, Mar 09, 2021 at 01:45:41PM +0100, Kornel Benko wrote:
> Am Tue, 9 Mar 2021 00:32:26 -0500
> schrieb Scott Kostyshak :
> 
> > > >
> > > >  $ egrep -i epub preferences
> > > > \format "" "epub" "Epub" "" "" "" "" ""
> > > > \converter "docbook5" "" "dbtoepub  -o $$o $$i" ""
> > > >
> > > > Restart lyx
> > > > remove converter docbook5 -> epub
> > > > Apply!
> > > > remove format epub
> > > > Apply! --> crash
> > > >
> > > >  $ egrep -i epub preferences
> > > > \format "" "epub" "Epub" "" "" "" "" ""
> > > >
> > > > The converter is gone, but the format still exists  
> > > 
> > > I still cannot reproduce this reliably. Probably best to file a bug.  
> > 
> > Did we file a bug report for this? I used to be able to reproduce. I
> > tried to reproduce on current master and could not, but I'm not sure if
> > I'm still following the recipe correctly.
> > 
> > Scott
> 
> Same here, cannot reproduce any more.
> To be sure, I even created new format (not contained in lyxrc.default), but 
> no crash.

OK good, thanks for checking.

Scott


signature.asc
Description: PGP signature
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2021-03-09 Thread Kornel Benko
Am Tue, 9 Mar 2021 00:32:26 -0500
schrieb Scott Kostyshak :

> > >
> > >  $ egrep -i epub preferences
> > > \format "" "epub" "Epub" "" "" "" "" ""
> > > \converter "docbook5" "" "dbtoepub  -o $$o $$i" ""
> > >
> > > Restart lyx
> > > remove converter docbook5 -> epub
> > > Apply!
> > > remove format epub
> > > Apply! --> crash
> > >
> > >  $ egrep -i epub preferences
> > > \format "" "epub" "Epub" "" "" "" "" ""
> > >
> > > The converter is gone, but the format still exists  
> > 
> > I still cannot reproduce this reliably. Probably best to file a bug.  
> 
> Did we file a bug report for this? I used to be able to reproduce. I
> tried to reproduce on current master and could not, but I'm not sure if
> I'm still following the recipe correctly.
> 
> Scott

Same here, cannot reproduce any more.
To be sure, I even created new format (not contained in lyxrc.default), but no 
crash.

Kornel


pgpTZeONsDAau.pgp
Description: Digitale Signatur von OpenPGP
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2021-03-08 Thread Scott Kostyshak
On Sat, Aug 08, 2020 at 12:39:50AM -0400, Richard Kimberly Heck wrote:
> On 7/6/20 12:32 PM, Kornel Benko wrote:
> > Am Mon, 6 Jul 2020 11:58:30 -0400
> > schrieb Richard Kimberly Heck :
> >
> >> On 7/6/20 4:50 AM, Kornel Benko wrote:
> >>> Am Sun, 5 Jul 2020 20:14:27 -0400
> >>> schrieb Richard Kimberly Heck :
> >>>
>  On 7/5/20 8:12 PM, Richard Kimberly Heck wrote:
> > On 6/24/20 11:06 AM, Richard Kimberly Heck wrote:  
> >> On 6/24/20 6:19 AM, Kornel Benko wrote:  
> >>> Am Tue, 23 Jun 2020 22:01:03 -0400
> >>> schrieb Scott Kostyshak :
> >>>  
>  On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:  
> > Am Mon, 22 Jun 2020 10:24:17 -0400
> > schrieb Richard Kimberly Heck :
> > 
> >> On 6/22/20 10:03 AM, Kornel Benko wrote:
> >>> Following lead to crash:
> >>>
> >>>   1. Create new file-Format (for instance Epub, define a viewer)
> >>>   2. define conversion for docbook-xls -> Epub
> >>>   3. restart lyx
> >>>   4. remove the new conversion
> >>>   5. remove the format Epub --> crash
> >>>
> >>> There is no crash if
> >>>   5. restart lyx
> >>>   6. remove the format Epub  
> >> I can't reproduce.
> > You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
> > -D_GLIBCXX_DEBUG"
> > 
> >> Can you post your preferences file as it exists after
> >> (3)?
> > 
> >> Do you save or apply after (4) or go right to step (5)?
> > Apply only. But with save it is the same.
>  I can reproduce. In addition I get the following:
> 
>    Error: attempt to subscript container with out-of-bounds index 
>  21973, but 
>    container only holds 99 elements.
> 
>  Scott  
> >>> Here the message is:
> >>> /usr/include/c++/8/debug/vector:417:
> >>> Error: attempt to subscript container with out-of-bounds index 
> >>> 1899910496, 
> >>> but container only holds 98 elements.
> >>>
> >>> Objects involved in the operation:
> >>> sequence "this" @ 0x0x555f98a58bc8 {
> >>>   type = std::__debug::vector >>> std::allocator  
>  ; }  
> >>> Abort (core dumped)
> >>>
> >>> If using attached, I got no crash, but the list of indexes used while 
> >>> removing the
> >>> format is scarifying.
> >>>
> >>>  $ wc errors
> >>>  1014  5060 35805 errors
> >>>
> >>>  $ sort my_list | uniq
> >>> Graph.cpp (145): VERT-INDEX == 0
> >>> Graph.cpp (145): VERT-INDEX == 11
> >>> Graph.cpp (145): VERT-INDEX == 21972
> >>> Graph.cpp (145): VERT-INDEX == 24
> >>> Graph.cpp (145): VERT-INDEX == 32
> >>> Graph.cpp (145): VERT-INDEX == 32757
> >>> Graph.cpp (145): VERT-INDEX == 33
> >>> Graph.cpp (145): VERT-INDEX == 71
> >>> Graph.cpp (145): VERT-INDEX == 79
> >>> Graph.cpp (145): VERT-INDEX == 8
> >>> Graph.cpp (145): VERT-INDEX == 80
> >>> Graph.cpp (145): VERT-INDEX == 83
> >>>
> >>>  $ egrep 21972 my_list | wc
> >>> 3771885   13949
> >>>
> >>>  $ egrep 32757 my_list | wc
> >>> 134 6704958  
> >> I will have a look.  
> > I was able to reproduce with the cmake setting mentioned above, but now
> > I can't again. I have not been able to reproduce without those settings,
> > nor do I see the weird values. But something is obviously suspicious.
> >
> > When I did see the crash, it was coming from Converters::isReachable().
> > You could try something like the attached, which could yield more
> > information.  
>  Sorry, changed wrong routine. Try this patch.
> 
>  Riki
> 
> 
> >>> To achieve the settings with gnu c++ compiler do:
> >>> with cmake: -DLYX_STDLIB_DEBUG=ON
> >>> with automake: --enable_stdlib_debug=yes
> >> I still can't get it to crash.
> >>
> >> Attached are the cmake settings I'm using. (This is generated by 
> >> QtCreator.)
> >>
> >> Riki
> >>
> >>
> > Looks good.
> >
> > Step1 create forma epub and converter docbook5 -> epub
> >
> >  $ egrep -i epub preferences
> > \format "" "epub" "Epub" "" "" "" "" ""
> > \converter "docbook5" "" "dbtoepub  -o $$o $$i" ""
> >
> > Restart lyx
> > remove converter docbook5 -> epub
> > Apply!
> > remove format epub
> > Apply! --> crash
> >
> >  $ egrep -i epub preferences
> > \format "" "epub" "Epub" "" "" "" "" ""
> >
> > The converter is gone, but the format still exists
> 
> I still cannot reproduce this reliably. Probably best to file a bug.

Did we file a bug report for this? I used to be able to reproduce. I
tried to reproduce on current master and could not, but I'm not sure if
I'm still following the recipe correctly.

Scott


signature.asc
Description: PGP signature
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org

Re: Crash removing Format

2020-08-07 Thread Richard Kimberly Heck
On 7/6/20 12:32 PM, Kornel Benko wrote:
> Am Mon, 6 Jul 2020 11:58:30 -0400
> schrieb Richard Kimberly Heck :
>
>> On 7/6/20 4:50 AM, Kornel Benko wrote:
>>> Am Sun, 5 Jul 2020 20:14:27 -0400
>>> schrieb Richard Kimberly Heck :
>>>
 On 7/5/20 8:12 PM, Richard Kimberly Heck wrote:
> On 6/24/20 11:06 AM, Richard Kimberly Heck wrote:  
>> On 6/24/20 6:19 AM, Kornel Benko wrote:  
>>> Am Tue, 23 Jun 2020 22:01:03 -0400
>>> schrieb Scott Kostyshak :
>>>  
 On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:  
> Am Mon, 22 Jun 2020 10:24:17 -0400
> schrieb Richard Kimberly Heck :
> 
>> On 6/22/20 10:03 AM, Kornel Benko wrote:
>>> Following lead to crash:
>>>
>>> 1. Create new file-Format (for instance Epub, define a viewer)
>>> 2. define conversion for docbook-xls -> Epub
>>> 3. restart lyx
>>> 4. remove the new conversion
>>> 5. remove the format Epub --> crash
>>>
>>> There is no crash if
>>> 5. restart lyx
>>> 6. remove the format Epub  
>> I can't reproduce.
> You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
> -D_GLIBCXX_DEBUG"
> 
>> Can you post your preferences file as it exists after
>> (3)?
> 
>> Do you save or apply after (4) or go right to step (5)?
> Apply only. But with save it is the same.
 I can reproduce. In addition I get the following:

   Error: attempt to subscript container with out-of-bounds index 
 21973, but 
   container only holds 99 elements.

 Scott  
>>> Here the message is:
>>> /usr/include/c++/8/debug/vector:417:
>>> Error: attempt to subscript container with out-of-bounds index 
>>> 1899910496, 
>>> but container only holds 98 elements.
>>>
>>> Objects involved in the operation:
>>> sequence "this" @ 0x0x555f98a58bc8 {
>>>   type = std::__debug::vector>> std::allocator  
 ; }  
>>> Abort (core dumped)
>>>
>>> If using attached, I got no crash, but the list of indexes used while 
>>> removing the
>>> format is scarifying.
>>>
>>>  $ wc errors
>>>  1014  5060 35805 errors
>>>
>>>  $ sort my_list | uniq
>>> Graph.cpp (145): VERT-INDEX == 0
>>> Graph.cpp (145): VERT-INDEX == 11
>>> Graph.cpp (145): VERT-INDEX == 21972
>>> Graph.cpp (145): VERT-INDEX == 24
>>> Graph.cpp (145): VERT-INDEX == 32
>>> Graph.cpp (145): VERT-INDEX == 32757
>>> Graph.cpp (145): VERT-INDEX == 33
>>> Graph.cpp (145): VERT-INDEX == 71
>>> Graph.cpp (145): VERT-INDEX == 79
>>> Graph.cpp (145): VERT-INDEX == 8
>>> Graph.cpp (145): VERT-INDEX == 80
>>> Graph.cpp (145): VERT-INDEX == 83
>>>
>>>  $ egrep 21972 my_list | wc
>>> 3771885   13949
>>>
>>>  $ egrep 32757 my_list | wc
>>> 134 6704958  
>> I will have a look.  
> I was able to reproduce with the cmake setting mentioned above, but now
> I can't again. I have not been able to reproduce without those settings,
> nor do I see the weird values. But something is obviously suspicious.
>
> When I did see the crash, it was coming from Converters::isReachable().
> You could try something like the attached, which could yield more
> information.  
 Sorry, changed wrong routine. Try this patch.

 Riki


>>> To achieve the settings with gnu c++ compiler do:
>>> with cmake: -DLYX_STDLIB_DEBUG=ON
>>> with automake: --enable_stdlib_debug=yes
>> I still can't get it to crash.
>>
>> Attached are the cmake settings I'm using. (This is generated by QtCreator.)
>>
>> Riki
>>
>>
> Looks good.
>
> Step1 create forma epub and converter docbook5 -> epub
>
>  $ egrep -i epub preferences
> \format "" "epub" "Epub" "" "" "" "" ""
> \converter "docbook5" "" "dbtoepub  -o $$o $$i" ""
>
> Restart lyx
> remove converter docbook5 -> epub
> Apply!
> remove format epub
> Apply! --> crash
>
>  $ egrep -i epub preferences
> \format "" "epub" "Epub" "" "" "" "" ""
>
> The converter is gone, but the format still exists

I still cannot reproduce this reliably. Probably best to file a bug.

Riki


-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-07-06 Thread Kornel Benko
Am Mon, 6 Jul 2020 11:58:30 -0400
schrieb Richard Kimberly Heck :

> On 7/6/20 4:50 AM, Kornel Benko wrote:
> > Am Sun, 5 Jul 2020 20:14:27 -0400
> > schrieb Richard Kimberly Heck :
> >
> >> On 7/5/20 8:12 PM, Richard Kimberly Heck wrote:
> >>> On 6/24/20 11:06 AM, Richard Kimberly Heck wrote:  
>  On 6/24/20 6:19 AM, Kornel Benko wrote:  
> > Am Tue, 23 Jun 2020 22:01:03 -0400
> > schrieb Scott Kostyshak :
> >  
> >> On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:  
> >>> Am Mon, 22 Jun 2020 10:24:17 -0400
> >>> schrieb Richard Kimberly Heck :
> >>> 
>  On 6/22/20 10:03 AM, Kornel Benko wrote:
> > Following lead to crash:
> >
> > 1. Create new file-Format (for instance Epub, define a viewer)
> > 2. define conversion for docbook-xls -> Epub
> > 3. restart lyx
> > 4. remove the new conversion
> > 5. remove the format Epub --> crash
> >
> > There is no crash if
> > 5. restart lyx
> > 6. remove the format Epub  
>  I can't reproduce.
> >>> You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
> >>> -D_GLIBCXX_DEBUG"
> >>> 
>  Can you post your preferences file as it exists after
>  (3)?
> >>> 
>  Do you save or apply after (4) or go right to step (5)?
> >>> Apply only. But with save it is the same.
> >> I can reproduce. In addition I get the following:
> >>
> >>   Error: attempt to subscript container with out-of-bounds index 
> >> 21973, but 
> >>   container only holds 99 elements.
> >>
> >> Scott  
> > Here the message is:
> > /usr/include/c++/8/debug/vector:417:
> > Error: attempt to subscript container with out-of-bounds index 
> > 1899910496, 
> > but container only holds 98 elements.
> >
> > Objects involved in the operation:
> > sequence "this" @ 0x0x555f98a58bc8 {
> >   type = std::__debug::vector > std::allocator  
> >> ; }  
> > Abort (core dumped)
> >
> > If using attached, I got no crash, but the list of indexes used while 
> > removing the
> > format is scarifying.
> >
> >  $ wc errors
> >  1014  5060 35805 errors
> >
> >  $ sort my_list | uniq
> > Graph.cpp (145): VERT-INDEX == 0
> > Graph.cpp (145): VERT-INDEX == 11
> > Graph.cpp (145): VERT-INDEX == 21972
> > Graph.cpp (145): VERT-INDEX == 24
> > Graph.cpp (145): VERT-INDEX == 32
> > Graph.cpp (145): VERT-INDEX == 32757
> > Graph.cpp (145): VERT-INDEX == 33
> > Graph.cpp (145): VERT-INDEX == 71
> > Graph.cpp (145): VERT-INDEX == 79
> > Graph.cpp (145): VERT-INDEX == 8
> > Graph.cpp (145): VERT-INDEX == 80
> > Graph.cpp (145): VERT-INDEX == 83
> >
> >  $ egrep 21972 my_list | wc
> > 3771885   13949
> >
> >  $ egrep 32757 my_list | wc
> > 134 6704958  
>  I will have a look.  
> >>> I was able to reproduce with the cmake setting mentioned above, but now
> >>> I can't again. I have not been able to reproduce without those settings,
> >>> nor do I see the weird values. But something is obviously suspicious.
> >>>
> >>> When I did see the crash, it was coming from Converters::isReachable().
> >>> You could try something like the attached, which could yield more
> >>> information.  
> >> Sorry, changed wrong routine. Try this patch.
> >>
> >> Riki
> >>
> >>
> > To achieve the settings with gnu c++ compiler do:
> > with cmake: -DLYX_STDLIB_DEBUG=ON
> > with automake: --enable_stdlib_debug=yes
> 
> I still can't get it to crash.
> 
> Attached are the cmake settings I'm using. (This is generated by QtCreator.)
> 
> Riki
> 
> 

Looks good.

Step1 create forma epub and converter docbook5 -> epub

 $ egrep -i epub preferences
\format "" "epub" "Epub" "" "" "" "" ""
\converter "docbook5" "" "dbtoepub  -o $$o $$i" ""

Restart lyx
remove converter docbook5 -> epub
Apply!
remove format epub
Apply! --> crash

 $ egrep -i epub preferences
\format "" "epub" "Epub" "" "" "" "" ""

The converter is gone, but the format still exists


Kornel


pgpoknlDo6Kn3.pgp
Description: Digitale Signatur von OpenPGP
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-07-06 Thread Richard Kimberly Heck
On 7/6/20 4:50 AM, Kornel Benko wrote:
> Am Sun, 5 Jul 2020 20:14:27 -0400
> schrieb Richard Kimberly Heck :
>
>> On 7/5/20 8:12 PM, Richard Kimberly Heck wrote:
>>> On 6/24/20 11:06 AM, Richard Kimberly Heck wrote:  
 On 6/24/20 6:19 AM, Kornel Benko wrote:  
> Am Tue, 23 Jun 2020 22:01:03 -0400
> schrieb Scott Kostyshak :
>  
>> On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:  
>>> Am Mon, 22 Jun 2020 10:24:17 -0400
>>> schrieb Richard Kimberly Heck :
>>> 
 On 6/22/20 10:03 AM, Kornel Benko wrote:
> Following lead to crash:
>
>   1. Create new file-Format (for instance Epub, define a viewer)
>   2. define conversion for docbook-xls -> Epub
>   3. restart lyx
>   4. remove the new conversion
>   5. remove the format Epub --> crash
>
> There is no crash if
>   5. restart lyx
>   6. remove the format Epub  
 I can't reproduce.
>>> You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
>>> -D_GLIBCXX_DEBUG"
>>> 
 Can you post your preferences file as it exists after
 (3)?
>>> 
 Do you save or apply after (4) or go right to step (5)?
>>> Apply only. But with save it is the same.
>> I can reproduce. In addition I get the following:
>>
>>   Error: attempt to subscript container with out-of-bounds index 21973, 
>> but 
>>   container only holds 99 elements.
>>
>> Scott  
> Here the message is:
> /usr/include/c++/8/debug/vector:417:
> Error: attempt to subscript container with out-of-bounds index 
> 1899910496, 
> but container only holds 98 elements.
>
> Objects involved in the operation:
> sequence "this" @ 0x0x555f98a58bc8 {
>   type = std::__debug::vector std::allocator  
>> ; }  
> Abort (core dumped)
>
> If using attached, I got no crash, but the list of indexes used while 
> removing the
> format is scarifying.
>
>  $ wc errors
>  1014  5060 35805 errors
>
>  $ sort my_list | uniq
> Graph.cpp (145): VERT-INDEX == 0
> Graph.cpp (145): VERT-INDEX == 11
> Graph.cpp (145): VERT-INDEX == 21972
> Graph.cpp (145): VERT-INDEX == 24
> Graph.cpp (145): VERT-INDEX == 32
> Graph.cpp (145): VERT-INDEX == 32757
> Graph.cpp (145): VERT-INDEX == 33
> Graph.cpp (145): VERT-INDEX == 71
> Graph.cpp (145): VERT-INDEX == 79
> Graph.cpp (145): VERT-INDEX == 8
> Graph.cpp (145): VERT-INDEX == 80
> Graph.cpp (145): VERT-INDEX == 83
>
>  $ egrep 21972 my_list | wc
> 3771885   13949
>
>  $ egrep 32757 my_list | wc
> 134 6704958  
 I will have a look.  
>>> I was able to reproduce with the cmake setting mentioned above, but now
>>> I can't again. I have not been able to reproduce without those settings,
>>> nor do I see the weird values. But something is obviously suspicious.
>>>
>>> When I did see the crash, it was coming from Converters::isReachable().
>>> You could try something like the attached, which could yield more
>>> information.  
>> Sorry, changed wrong routine. Try this patch.
>>
>> Riki
>>
>>
> To achieve the settings with gnu c++ compiler do:
> with cmake: -DLYX_STDLIB_DEBUG=ON
> with automake: --enable_stdlib_debug=yes

I still can't get it to crash.

Attached are the cmake settings I'm using. (This is generated by QtCreator.)

Riki




run_cmake.sh
Description: application/shellscript
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-07-06 Thread Kornel Benko
Am Mon, 6 Jul 2020 09:51:07 -0400
schrieb Scott Kostyshak :

> On Mon, Jul 06, 2020 at 10:50:10AM +0200, Kornel Benko wrote:
> 
> > with automake: --enable_stdlib_debug=yes  
> 
> I think the underscores need to be dashes:
> 
>   --enable-stdlib-debug=yes
> 
> Scott

Right, copy & paste from configure script.

Kornel


pgpunVB_qCfsa.pgp
Description: Digitale Signatur von OpenPGP
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-07-06 Thread Scott Kostyshak
On Mon, Jul 06, 2020 at 10:50:10AM +0200, Kornel Benko wrote:

> with automake: --enable_stdlib_debug=yes

I think the underscores need to be dashes:

  --enable-stdlib-debug=yes

Scott


signature.asc
Description: PGP signature
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-07-06 Thread Kornel Benko
Am Sun, 5 Jul 2020 20:14:27 -0400
schrieb Richard Kimberly Heck :

> On 7/5/20 8:12 PM, Richard Kimberly Heck wrote:
> > On 6/24/20 11:06 AM, Richard Kimberly Heck wrote:  
> >> On 6/24/20 6:19 AM, Kornel Benko wrote:  
> >>> Am Tue, 23 Jun 2020 22:01:03 -0400
> >>> schrieb Scott Kostyshak :
> >>>  
>  On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:  
> > Am Mon, 22 Jun 2020 10:24:17 -0400
> > schrieb Richard Kimberly Heck :
> > 
> >> On 6/22/20 10:03 AM, Kornel Benko wrote:
> >>> Following lead to crash:
> >>>
> >>>   1. Create new file-Format (for instance Epub, define a viewer)
> >>>   2. define conversion for docbook-xls -> Epub
> >>>   3. restart lyx
> >>>   4. remove the new conversion
> >>>   5. remove the format Epub --> crash
> >>>
> >>> There is no crash if
> >>>   5. restart lyx
> >>>   6. remove the format Epub  
> >> I can't reproduce.
> > You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
> > -D_GLIBCXX_DEBUG"
> > 
> >> Can you post your preferences file as it exists after
> >> (3)?
> > 
> >> Do you save or apply after (4) or go right to step (5)?
> > Apply only. But with save it is the same.
>  I can reproduce. In addition I get the following:
> 
>    Error: attempt to subscript container with out-of-bounds index 21973, 
>  but 
>    container only holds 99 elements.
> 
>  Scott  
> >>> Here the message is:
> >>> /usr/include/c++/8/debug/vector:417:
> >>> Error: attempt to subscript container with out-of-bounds index 
> >>> 1899910496, 
> >>> but container only holds 98 elements.
> >>>
> >>> Objects involved in the operation:
> >>> sequence "this" @ 0x0x555f98a58bc8 {
> >>>   type = std::__debug::vector >>> std::allocator  
>  ; }  
> >>> Abort (core dumped)
> >>>
> >>> If using attached, I got no crash, but the list of indexes used while 
> >>> removing the
> >>> format is scarifying.
> >>>
> >>>  $ wc errors
> >>>  1014  5060 35805 errors
> >>>
> >>>  $ sort my_list | uniq
> >>> Graph.cpp (145): VERT-INDEX == 0
> >>> Graph.cpp (145): VERT-INDEX == 11
> >>> Graph.cpp (145): VERT-INDEX == 21972
> >>> Graph.cpp (145): VERT-INDEX == 24
> >>> Graph.cpp (145): VERT-INDEX == 32
> >>> Graph.cpp (145): VERT-INDEX == 32757
> >>> Graph.cpp (145): VERT-INDEX == 33
> >>> Graph.cpp (145): VERT-INDEX == 71
> >>> Graph.cpp (145): VERT-INDEX == 79
> >>> Graph.cpp (145): VERT-INDEX == 8
> >>> Graph.cpp (145): VERT-INDEX == 80
> >>> Graph.cpp (145): VERT-INDEX == 83
> >>>
> >>>  $ egrep 21972 my_list | wc
> >>> 3771885   13949
> >>>
> >>>  $ egrep 32757 my_list | wc
> >>> 134 6704958  
> >> I will have a look.  
> > I was able to reproduce with the cmake setting mentioned above, but now
> > I can't again. I have not been able to reproduce without those settings,
> > nor do I see the weird values. But something is obviously suspicious.
> >
> > When I did see the crash, it was coming from Converters::isReachable().
> > You could try something like the attached, which could yield more
> > information.  
> 
> Sorry, changed wrong routine. Try this patch.
> 
> Riki
> 
> 

To achieve the settings with gnu c++ compiler do:
with cmake: -DLYX_STDLIB_DEBUG=ON
with automake: --enable_stdlib_debug=yes

With your patch I get:
...
Converter.cpp (987): From pdflatex==34 to xpm==20
Converter.cpp (987): From platex==0 to xpm==20
Converter.cpp (987): From xhtml==50 to bmp==1
/usr/include/c++/8/debug/vector:417:
Error: attempt to subscript container with out-of-bounds index 21870, but 
container only holds 98 elements.

Objects involved in the operation:
sequence "this" @ 0x0x556e87b6f4b8 {
  type = std::__debug::vector
>;
}
Abort (core dumped)

...
#0  __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
#1  0x74c94801 in __GI_abort () at abort.c:79
#2  0x75689789 in ?? () from /usr/lib/x86_64-linux-gnu/libstdc++.so.6
#3  0x565d53ed in std::__debug::vector >::operator[] (this=0x59d7fdb8, 
__n=21845)
at /usr/include/c++/8/debug/vector:417
#4  0x565d4168 in lyx::Graph::isReachable (this=0x59d7fd88, 
from=1, to=49) at /usr2/src/lyx/lyx-git/src/Graph.cpp:145
#5  0x5651dd0d in lyx::Converters::isReachable (this=0x59d7fcb0, 
from="xhtml", to="bmp") at /usr2/src/lyx/lyx-git/src/Converter.cpp:989
#6  0x56d40343 in lyx::frontend::PrefFileformats::updateView (
this=0x581a9620)
at /usr2/src/lyx/lyx-git/src/frontends/qt/GuiPrefs.cpp:2090

(This was the try to remove file format docbook-xml.)

Kornel


pgp5YAvBSBx8R.pgp
Description: Digitale Signatur von OpenPGP
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-07-05 Thread Richard Kimberly Heck
On 7/5/20 8:12 PM, Richard Kimberly Heck wrote:
> On 6/24/20 11:06 AM, Richard Kimberly Heck wrote:
>> On 6/24/20 6:19 AM, Kornel Benko wrote:
>>> Am Tue, 23 Jun 2020 22:01:03 -0400
>>> schrieb Scott Kostyshak :
>>>
 On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:
> Am Mon, 22 Jun 2020 10:24:17 -0400
> schrieb Richard Kimberly Heck :
>   
>> On 6/22/20 10:03 AM, Kornel Benko wrote:  
>>> Following lead to crash:
>>>
>>> 1. Create new file-Format (for instance Epub, define a viewer)
>>> 2. define conversion for docbook-xls -> Epub
>>> 3. restart lyx
>>> 4. remove the new conversion
>>> 5. remove the format Epub --> crash
>>>
>>> There is no crash if
>>> 5. restart lyx
>>> 6. remove the format Epub
>> I can't reproduce.  
> You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
> -D_GLIBCXX_DEBUG"
>   
>> Can you post your preferences file as it exists after
>> (3)?  
>   
>> Do you save or apply after (4) or go right to step (5)?  
> Apply only. But with save it is the same.  
 I can reproduce. In addition I get the following:

   Error: attempt to subscript container with out-of-bounds index 21973, 
 but 
   container only holds 99 elements.

 Scott
>>> Here the message is:
>>> /usr/include/c++/8/debug/vector:417:
>>> Error: attempt to subscript container with out-of-bounds index 1899910496, 
>>> but container only holds 98 elements.
>>>
>>> Objects involved in the operation:
>>> sequence "this" @ 0x0x555f98a58bc8 {
>>>   type = std::__debug::vector>> std::allocator
 ; }
>>> Abort (core dumped)
>>>
>>> If using attached, I got no crash, but the list of indexes used while 
>>> removing the format
>>> is scarifying.
>>>
>>>  $ wc errors
>>>  1014  5060 35805 errors
>>>
>>>  $ sort my_list | uniq
>>> Graph.cpp (145): VERT-INDEX == 0
>>> Graph.cpp (145): VERT-INDEX == 11
>>> Graph.cpp (145): VERT-INDEX == 21972
>>> Graph.cpp (145): VERT-INDEX == 24
>>> Graph.cpp (145): VERT-INDEX == 32
>>> Graph.cpp (145): VERT-INDEX == 32757
>>> Graph.cpp (145): VERT-INDEX == 33
>>> Graph.cpp (145): VERT-INDEX == 71
>>> Graph.cpp (145): VERT-INDEX == 79
>>> Graph.cpp (145): VERT-INDEX == 8
>>> Graph.cpp (145): VERT-INDEX == 80
>>> Graph.cpp (145): VERT-INDEX == 83
>>>
>>>  $ egrep 21972 my_list | wc
>>> 3771885   13949
>>>
>>>  $ egrep 32757 my_list | wc
>>> 134 6704958
>> I will have a look.
> I was able to reproduce with the cmake setting mentioned above, but now
> I can't again. I have not been able to reproduce without those settings,
> nor do I see the weird values. But something is obviously suspicious.
>
> When I did see the crash, it was coming from Converters::isReachable().
> You could try something like the attached, which could yield more
> information.

Sorry, changed wrong routine. Try this patch.

Riki


diff --git a/src/Converter.cpp b/src/Converter.cpp
index 3487f4b45e..dc4be69cc3 100644
--- a/src/Converter.cpp
+++ b/src/Converter.cpp
@@ -982,8 +982,11 @@ FormatList const Converters::getReachable(string const & 
from,
 
 bool Converters::isReachable(string const & from, string const & to)
 {
-   return G_.isReachable(theFormats().getNumber(from),
- theFormats().getNumber(to));
+   int const f = theFormats().getNumber(from);
+   int const t = theFormats().getNumber(to);
+   LYXERR0("From " << from << "==" << f << " to " << to << "==" << t);
+   LATTEST(t < 1000 && f < 1000);
+   return G_.isReachable(t, f);
 }
 
 
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-07-05 Thread Richard Kimberly Heck
On 6/24/20 11:06 AM, Richard Kimberly Heck wrote:
> On 6/24/20 6:19 AM, Kornel Benko wrote:
>> Am Tue, 23 Jun 2020 22:01:03 -0400
>> schrieb Scott Kostyshak :
>>
>>> On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:
 Am Mon, 22 Jun 2020 10:24:17 -0400
 schrieb Richard Kimberly Heck :
   
> On 6/22/20 10:03 AM, Kornel Benko wrote:  
>> Following lead to crash:
>>
>>  1. Create new file-Format (for instance Epub, define a viewer)
>>  2. define conversion for docbook-xls -> Epub
>>  3. restart lyx
>>  4. remove the new conversion
>>  5. remove the format Epub --> crash
>>
>> There is no crash if
>>  5. restart lyx
>>  6. remove the format Epub
> I can't reproduce.  
 You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
 -D_GLIBCXX_DEBUG"
   
> Can you post your preferences file as it exists after
> (3)?  
   
> Do you save or apply after (4) or go right to step (5)?  
 Apply only. But with save it is the same.  
>>> I can reproduce. In addition I get the following:
>>>
>>>   Error: attempt to subscript container with out-of-bounds index 21973, but 
>>>   container only holds 99 elements.
>>>
>>> Scott
>> Here the message is:
>> /usr/include/c++/8/debug/vector:417:
>> Error: attempt to subscript container with out-of-bounds index 1899910496, 
>> but container only holds 98 elements.
>>
>> Objects involved in the operation:
>> sequence "this" @ 0x0x555f98a58bc8 {
>>   type = std::__debug::vector> std::allocator
>>> ; }
>> Abort (core dumped)
>>
>> If using attached, I got no crash, but the list of indexes used while 
>> removing the format
>> is scarifying.
>>
>>  $ wc errors
>>  1014  5060 35805 errors
>>
>>  $ sort my_list | uniq
>> Graph.cpp (145): VERT-INDEX == 0
>> Graph.cpp (145): VERT-INDEX == 11
>> Graph.cpp (145): VERT-INDEX == 21972
>> Graph.cpp (145): VERT-INDEX == 24
>> Graph.cpp (145): VERT-INDEX == 32
>> Graph.cpp (145): VERT-INDEX == 32757
>> Graph.cpp (145): VERT-INDEX == 33
>> Graph.cpp (145): VERT-INDEX == 71
>> Graph.cpp (145): VERT-INDEX == 79
>> Graph.cpp (145): VERT-INDEX == 8
>> Graph.cpp (145): VERT-INDEX == 80
>> Graph.cpp (145): VERT-INDEX == 83
>>
>>  $ egrep 21972 my_list | wc
>> 3771885   13949
>>
>>  $ egrep 32757 my_list | wc
>> 134 6704958
> I will have a look.

I was able to reproduce with the cmake setting mentioned above, but now
I can't again. I have not been able to reproduce without those settings,
nor do I see the weird values. But something is obviously suspicious.

When I did see the crash, it was coming from Converters::isReachable().
You could try something like the attached, which could yield more
information.

Riki



diff --git a/src/Converter.cpp b/src/Converter.cpp
index 3487f4b45e..172409873e 100644
--- a/src/Converter.cpp
+++ b/src/Converter.cpp
@@ -989,8 +989,11 @@ bool Converters::isReachable(string const & from, string 
const & to)
 
 Graph::EdgePath Converters::getPath(string const & from, string const & to)
 {
-   return G_.getPath(theFormats().getNumber(from),
- theFormats().getNumber(to));
+   int const f = theFormats().getNumber(from);
+   int const t = theFormats().getNumber(to);
+   LYXERR0("From " << from << "==" << f << " to " << to << "==" << t);
+   LATTEST(t < 1000 && f < 1000);
+   return G_.getPath(f, t);
 }
 
 
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-06-24 Thread Richard Kimberly Heck
On 6/24/20 6:19 AM, Kornel Benko wrote:
> Am Tue, 23 Jun 2020 22:01:03 -0400
> schrieb Scott Kostyshak :
>
>> On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:
>>> Am Mon, 22 Jun 2020 10:24:17 -0400
>>> schrieb Richard Kimberly Heck :
>>>   
 On 6/22/20 10:03 AM, Kornel Benko wrote:  
> Following lead to crash:
>
>   1. Create new file-Format (for instance Epub, define a viewer)
>   2. define conversion for docbook-xls -> Epub
>   3. restart lyx
>   4. remove the new conversion
>   5. remove the format Epub --> crash
>
> There is no crash if
>   5. restart lyx
>   6. remove the format Epub
 I can't reproduce.  
>>> You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
>>> -D_GLIBCXX_DEBUG"
>>>   
 Can you post your preferences file as it exists after
 (3)?  
>>>   
 Do you save or apply after (4) or go right to step (5)?  
>>> Apply only. But with save it is the same.  
>> I can reproduce. In addition I get the following:
>>
>>   Error: attempt to subscript container with out-of-bounds index 21973, but 
>>   container only holds 99 elements.
>>
>> Scott
> Here the message is:
> /usr/include/c++/8/debug/vector:417:
> Error: attempt to subscript container with out-of-bounds index 1899910496, 
> but container only holds 98 elements.
>
> Objects involved in the operation:
> sequence "this" @ 0x0x555f98a58bc8 {
>   type = std::__debug::vector std::allocator
>> ; }
> Abort (core dumped)
>
> If using attached, I got no crash, but the list of indexes used while 
> removing the format
> is scarifying.
>
>  $ wc errors
>  1014  5060 35805 errors
>
>  $ sort my_list | uniq
> Graph.cpp (145): VERT-INDEX == 0
> Graph.cpp (145): VERT-INDEX == 11
> Graph.cpp (145): VERT-INDEX == 21972
> Graph.cpp (145): VERT-INDEX == 24
> Graph.cpp (145): VERT-INDEX == 32
> Graph.cpp (145): VERT-INDEX == 32757
> Graph.cpp (145): VERT-INDEX == 33
> Graph.cpp (145): VERT-INDEX == 71
> Graph.cpp (145): VERT-INDEX == 79
> Graph.cpp (145): VERT-INDEX == 8
> Graph.cpp (145): VERT-INDEX == 80
> Graph.cpp (145): VERT-INDEX == 83
>
>  $ egrep 21972 my_list | wc
> 3771885   13949
>
>  $ egrep 32757 my_list | wc
> 134 6704958

I will have a look.

Riki


-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-06-24 Thread Kornel Benko
Am Tue, 23 Jun 2020 22:01:03 -0400
schrieb Scott Kostyshak :

> On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:
> > Am Mon, 22 Jun 2020 10:24:17 -0400
> > schrieb Richard Kimberly Heck :
> >   
> > > On 6/22/20 10:03 AM, Kornel Benko wrote:  
> > > > Following lead to crash:
> > > >
> > > > 1. Create new file-Format (for instance Epub, define a viewer)
> > > > 2. define conversion for docbook-xls -> Epub
> > > > 3. restart lyx
> > > > 4. remove the new conversion
> > > > 5. remove the format Epub --> crash
> > > >
> > > > There is no crash if
> > > > 5. restart lyx
> > > > 6. remove the format Epub
> > > 
> > > I can't reproduce.  
> > 
> > You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
> > -D_GLIBCXX_DEBUG"
> >   
> > > Can you post your preferences file as it exists after
> > > (3)?  
> > 
> >   
> > > Do you save or apply after (4) or go right to step (5)?  
> > 
> > Apply only. But with save it is the same.  
> 
> I can reproduce. In addition I get the following:
> 
>   Error: attempt to subscript container with out-of-bounds index 21973, but 
>   container only holds 99 elements.
> 
> Scott

Here the message is:
/usr/include/c++/8/debug/vector:417:
Error: attempt to subscript container with out-of-bounds index 1899910496, 
but container only holds 98 elements.

Objects involved in the operation:
sequence "this" @ 0x0x555f98a58bc8 {
  type = std::__debug::vector
>; }
Abort (core dumped)

If using attached, I got no crash, but the list of indexes used while removing 
the format
is scarifying.

 $ wc errors
 1014  5060 35805 errors

 $ sort my_list | uniq
Graph.cpp (145): VERT-INDEX == 0
Graph.cpp (145): VERT-INDEX == 11
Graph.cpp (145): VERT-INDEX == 21972
Graph.cpp (145): VERT-INDEX == 24
Graph.cpp (145): VERT-INDEX == 32
Graph.cpp (145): VERT-INDEX == 32757
Graph.cpp (145): VERT-INDEX == 33
Graph.cpp (145): VERT-INDEX == 71
Graph.cpp (145): VERT-INDEX == 79
Graph.cpp (145): VERT-INDEX == 8
Graph.cpp (145): VERT-INDEX == 80
Graph.cpp (145): VERT-INDEX == 83

 $ egrep 21972 my_list | wc
3771885   13949

 $ egrep 32757 my_list | wc
134 6704958

 Kornel
diff --git a/src/Graph.cpp b/src/Graph.cpp
index 036bb38857..8374d38e24 100644
--- a/src/Graph.cpp
+++ b/src/Graph.cpp
@@ -140,10 +140,14 @@ bool Graph::isReachable(int from, int to)
 			vertices_[current].out_arrows.begin();
 		vector::const_iterator end =
 			vertices_[current].out_arrows.end();
 		for (; cit != end; ++cit) {
 			int const cv = (*cit)->to;
+			LYXERR0("VERT-INDEX == " << cv);
+			if (cv < 0 || cv >= vertices_.size()) {
+			  continue;
+			}
 			if (!vertices_[cv].visited) {
 vertices_[cv].visited = true;
 Q.push(cv);
 			}
 		}


pgpbS8HHnTFMO.pgp
Description: Digitale Signatur von OpenPGP
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-06-23 Thread Scott Kostyshak
On Mon, Jun 22, 2020 at 04:55:50PM +0200, Kornel Benko wrote:
> Am Mon, 22 Jun 2020 10:24:17 -0400
> schrieb Richard Kimberly Heck :
> 
> > On 6/22/20 10:03 AM, Kornel Benko wrote:
> > > Following lead to crash:
> > >
> > >   1. Create new file-Format (for instance Epub, define a viewer)
> > >   2. define conversion for docbook-xls -> Epub
> > >   3. restart lyx
> > >   4. remove the new conversion
> > >   5. remove the format Epub --> crash
> > >
> > > There is no crash if
> > >   5. restart lyx
> > >   6. remove the format Epub  
> > 
> > I can't reproduce.
> 
> You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC 
> -D_GLIBCXX_DEBUG"
> 
> > Can you post your preferences file as it exists after
> > (3)?
> 
> 
> > Do you save or apply after (4) or go right to step (5)?
> 
> Apply only. But with save it is the same.

I can reproduce. In addition I get the following:

  Error: attempt to subscript container with out-of-bounds index 21973, but 
  container only holds 99 elements.

Scott


signature.asc
Description: PGP signature
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-06-22 Thread Kornel Benko
Am Mon, 22 Jun 2020 10:24:17 -0400
schrieb Richard Kimberly Heck :

> On 6/22/20 10:03 AM, Kornel Benko wrote:
> > Following lead to crash:
> >
> > 1. Create new file-Format (for instance Epub, define a viewer)
> > 2. define conversion for docbook-xls -> Epub
> > 3. restart lyx
> > 4. remove the new conversion
> > 5. remove the format Epub --> crash
> >
> > There is no crash if
> > 5. restart lyx
> > 6. remove the format Epub  
> 
> I can't reproduce.

You probably need to recompile with "-D_GLIBCXX_DEBUG_PEDANTIC -D_GLIBCXX_DEBUG"

> Can you post your preferences file as it exists after
> (3)?


> Do you save or apply after (4) or go right to step (5)?

Apply only. But with save it is the same.

> 
> Riki
> 

Kornel



pgppwQsi8us7A.pgp
Description: Digitale Signatur von OpenPGP
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-06-22 Thread Kornel Benko
Am Mon, 22 Jun 2020 10:24:17 -0400
schrieb Richard Kimberly Heck :

> On 6/22/20 10:03 AM, Kornel Benko wrote:
> > Following lead to crash:
> >
> > 1. Create new file-Format (for instance Epub, define a viewer)
> > 2. define conversion for docbook-xls -> Epub
> > 3. restart lyx
> > 4. remove the new conversion
> > 5. remove the format Epub --> crash
> >
> > There is no crash if
> > 5. restart lyx
> > 6. remove the format Epub  
> 
> I can't reproduce. Can you post your preferences file as it exists after
> (3)? Do you save or apply after (4) or go right to step (5)?
> 
> Riki
> 
> 

Is this not enough?
$ egrep -i epub preferences
\format "" "epub" "Epub" "" "" "" "" ""
\converter "docbook-xml" "" "dbtoepub $$i -o $$o" ""

Attaching also backtrace:

Kornel



backtrace
Description: Binary data


pgp1v55WYEBnR.pgp
Description: Digitale Signatur von OpenPGP
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Re: Crash removing Format

2020-06-22 Thread Richard Kimberly Heck
On 6/22/20 10:03 AM, Kornel Benko wrote:
> Following lead to crash:
>
>   1. Create new file-Format (for instance Epub, define a viewer)
>   2. define conversion for docbook-xls -> Epub
>   3. restart lyx
>   4. remove the new conversion
>   5. remove the format Epub --> crash
>
> There is no crash if
>   5. restart lyx
>   6. remove the format Epub

I can't reproduce. Can you post your preferences file as it exists after
(3)? Do you save or apply after (4) or go right to step (5)?

Riki


-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel


Crash removing Format

2020-06-22 Thread Kornel Benko
Following lead to crash:

1. Create new file-Format (for instance Epub, define a viewer)
2. define conversion for docbook-xls -> Epub
3. restart lyx
4. remove the new conversion
5. remove the format Epub --> crash

There is no crash if
5. restart lyx
6. remove the format Epub

Kornel



pgpwA2G5ZdOnZ.pgp
Description: Digitale Signatur von OpenPGP
-- 
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel