Hey,
So I think that this patch fixes a bunch of the extensions problems, many of them on win32. Can someone test it on Windows for me?
Thanks, Ted
Hi Ted,
I just tried. Unfortunately it did not work :(
Did you notice my changes in trunk? I submitted the attached patch to Bryce and Josh to incorporate in 0.46, but I think it is not necessary seeing you did some changes in notebook parameter aswell. Could you sort it out in trunk?
(Please note that your patch contains a lot of tab characters that we are trying to get rid of!)
Inkscape shows the popup with "function plotter" working, but then crashes with:
C:\inkscape046\inkscape>inkscape terminate called after throwing an instance of 'Glib::SpawnError'
This application has requested the Runtime to terminate it in an unusual way. Please contact the application's support team for more information.
Emergency save activated! Emergency save completed. Inkscape will close now. If you can reproduce this crash, please file a bug at www.inkscape.org with a detailed description of the steps leading to the crash, so we can fix it.
Gdb tells me it crashes in a windows routine that is string related. Can't say where in Inkscape code though.
Inkscape-trunk now does *not* crash, however the function plotter still does not work. Another effect that does use the notebook parameter (measure path) *does* work. So that's good. I do get these warnings though:
** (inkscape.exe:18676): CRITICAL **: void Inkscape::GC::Anchored::release() const: assertion `_anchor' failed
Regards, Johan
-----Original Message----- From: inkscape-devel-bounces@lists.sourceforge.net [mailto:inkscape-devel-bounces@lists.sourceforge.net] On Behalf Of Ted Gould Sent: donderdag 28 februari 2008 6:08 To: Inkscape Devel List Subject: [Inkscape-devel] Extensions on win32
Hey,
So I think that this patch fixes a bunch of the extensions problems, many of them on win32. Can someone test it on Windows for me?
Thanks, Ted
On Thu, 2008-02-28 at 17:05 +0100, J.B.C.Engelen@...1578... wrote:
I just tried. Unfortunately it did not work :(
Hmm, so the live effects didn't work or the notebook changes didn't?
Did you notice my changes in trunk? I submitted the attached patch to Bryce and Josh to incorporate in 0.46, but I think it is not necessary seeing you did some changes in notebook parameter aswell. Could you sort it out in trunk?
Yes I will. In my personal opinion I like my patch better :) The big reason for me is that it moves all the "--<name>=" code to one place so that it's unlikely we'll have this bug again :)
(Please note that your patch contains a lot of tab characters that we are trying to get rid of!)
Err. Okay. I'll fix that.
Inkscape shows the popup with "function plotter" working, but then crashes with:
Aaron, can you get the list of parameters passed again? I think they should be correct now...
--Ted
Hi Ted,
Sorry for the short message 'it did not work'. I won't be able to mail soon, so just this quick reply. Since I am on different PC atm, I can't test what does and what does not work. I don't like my fix much either, yours is cleaner, so go ahead! Mine was more of a quick hack to get it working.
For some reason, the function plotter does work for me now, most of the time. Sometimes it bugs, don't know why, but it crashes. The GC::Anchored error message is always present though.
Thanks, Johan
-----Original Message----- From: Ted Gould [mailto:ted@...11...] Sent: Fri 2/29/2008 20:42 To: Engelen, J.B.C. (Johan); Aaron Spike Cc: inkscape-devel@lists.sourceforge.net Subject: RE: [Inkscape-devel] Extensions on win32
On Thu, 2008-02-28 at 17:05 +0100, J.B.C.Engelen@...1578... wrote:
I just tried. Unfortunately it did not work :(
Hmm, so the live effects didn't work or the notebook changes didn't?
Did you notice my changes in trunk? I submitted the attached patch to Bryce and Josh to incorporate in 0.46, but I think it is not necessary seeing you did some changes in notebook parameter aswell. Could you sort it out in trunk?
Yes I will. In my personal opinion I like my patch better :) The big reason for me is that it moves all the "--<name>=" code to one place so that it's unlikely we'll have this bug again :)
(Please note that your patch contains a lot of tab characters that we are trying to get rid of!)
Err. Okay. I'll fix that.
Inkscape shows the popup with "function plotter" working, but then crashes with:
Aaron, can you get the list of parameters passed again? I think they should be correct now...
--Ted
On Fri, 2008-02-29 at 22:31 +0100, J.B.C.Engelen@...1578... wrote:
For some reason, the function plotter does work for me now, most of the time. Sometimes it bugs, don't know why, but it crashes. The GC::Anchored error message is always present though.
Mental,
Is there a way I can use the logging mechanism to find out more about this. I'm getting it also on some of the effects, I think only scripts, I'm unsure why.
--Ted
On Fri, 2008-02-29 at 13:40 -0800, Ted Gould wrote:
On Fri, 2008-02-29 at 22:31 +0100, J.B.C.Engelen@...1578... wrote:
For some reason, the function plotter does work for me now, most of the time. Sometimes it bugs, don't know why, but it crashes. The GC::Anchored error message is always present though.
Is there a way I can use the logging mechanism to find out more about this. I'm getting it also on some of the effects, I think only scripts, I'm unsure why.
It sounds like the problem is at least partly unbalanced anchor counts.
Try setting the following environment variables accordingly and looking at the debug output.
INKSCAPE_DEBUG_LOG=debug.xml INKSCAPE_DEBUG_FILTER=REFCOUNT,EXTENSION,DOCUMENT
-mental
Ted Gould wrote:
On Fri, 2008-02-29 at 22:31 +0100, J.B.C.Engelen@...1578... wrote:
For some reason, the function plotter does work for me now, most of the time. Sometimes it bugs, don't know why, but it crashes. The GC::Anchored error message is always present though.
Mental,
Is there a way I can use the logging mechanism to find out more about this. I'm getting it also on some of the effects, I think only scripts, I'm unsure why.
--Ted
Ted,
Everyone should be using the g_something() style of logging messages, because glib allows us to hook into it and grab copies of the msgs. Anyone who does an assert() or printf() needs to be shot. :-)
But IF they do use the g_* messages, then we can grab and display them in the Messages dialog.
bob
Ted Gould schrieb:
On Fri, 2008-02-29 at 22:31 +0100, J.B.C.Engelen@...1578... wrote:
For some reason, the function plotter does work for me now, most of the time. Sometimes it bugs, don't know why, but it crashes. The GC::Anchored error message is always present though.
Mental,
Is there a way I can use the logging mechanism to find out more about this. I'm getting it also on some of the effects, I think only scripts, I'm unsure why.
--Ted
Hi,
I'm having a input extension which works as external application. It works fine with inkscape 0.45.1 but doesn't work with inkscape 0.46 preX. My application generate SVG to Stdout. If I use it in a windows shell and open the result in Inkscape 0.46pr2 and daily build it works, but no more as extension (inkscape crashes at this point). Is that an issue or are here changes in the configuration for external extensions?
Regards, Simon
Simon Mieth wrote:
I'm having a input extension which works as external application. It works fine with inkscape 0.45.1 but doesn't work with inkscape 0.46 preX. My application generate SVG to Stdout. If I use it in a windows shell and open the result in Inkscape 0.46pr2 and daily build it works, but no more as extension (inkscape crashes at this point). Is that an issue or are here changes in the configuration for external extensions?
We should probably start a new thread for this.
We need more information. You could be seeing one of two possible problems. A week or so ago I committed a one line fix to a bug that made extensions fail if they passed any binary data back to Inkscape on stdout. On vista there is an issue with the python environment that scripts see. This causes all python effects to fail.
Aaron
Aaron Spike schrieb:
Simon Mieth wrote:
I'm having a input extension which works as external application. It works fine with inkscape 0.45.1 but doesn't work with inkscape 0.46 preX. My application generate SVG to Stdout. If I use it in a windows shell and open the result in Inkscape 0.46pr2 and daily build it works, but no more as extension (inkscape crashes at this point). Is that an issue or are here changes in the configuration for external extensions?
We should probably start a new thread for this.
We need more information. You could be seeing one of two possible problems. A week or so ago I committed a one line fix to a bug that made extensions fail if they passed any binary data back to Inkscape on stdout. On vista there is an issue with the python environment that scripts see. This causes all python effects to fail.
Aaron
Hi Aaron,
I have a Java project which provides an API for parsing DXF and one part for converting this to SVG. The project includes a CLI tool like the inkscape-dxf2svg tool. It takes the file and output all SVG to stdout. If I use "mydxf2svg c:\path\to\dxf > result.svg" in Windows shell I can open the result in inkscape. With inkscape 0.45.1 and earlier it works with 0.46X not. I'm working for a new release and want just to know if there is something changed so that have to change my tool or is that an issue. The extension can be build as java extension or as .NET extension and both run in a Windows shell so it should not be a path issue. Windows XP SP2, Inkscape daily build from Feb 29 2008: Java extension and .NET error message
Microsoft Visual C++ Runtime Library Runtime Error! Inkscape internal error
My INX file: <inkscape-extension> <_name>Kabeja DXF Input</_name> <id>org.kabeja.inkscape.import.dxf</id> <dependency type="extension">org.inkscape.input.svg</dependency> <param name="description" type="description">Choose the layout for the import.</param> <param name="bounds-rule" gui-text="Layout" type="enum"> <item>Modelspace</item> <item>Modelspace-Limits</item> <item>Paperspace</item> <item>Paperspace-Limits</item> </param> <input> <extension>.dxf</extension> <mimetype>application/dxf</mimetype> <_filetypename>Kabeja-AutoCAD DXF (*.dxf)</_filetypename> <_filetypetooltip>Import AutoCAD's Document Exchange Format</_filetypetooltip> <output_extension>org.inkscape.output.svg</output_extension> </input> <script> <command reldir="extensions">kabeja-dxf2svg</command> </script> </inkscape-extension>
So maybe something goes wrong with the pipe to inkscape.
Regards, Simon
Ted Gould wrote:
Aaron, can you get the list of parameters passed again? I think they should be correct now...
With r17570:
['share\extensions\funcplot.py', 'C:\DOCUME~1\NS\LOCALS~1\Temp\ink_ext_XXXXXX.svgTDGF7T', '--id=rect3171', '--tab="sampling"', '--xstart=0', '--xend=10', '--times2pi=true', '--ybottom=-11.600000381469727', '--ytop=24.600000381469727', '--samples=194', '--isoscale=false', '--polar=false', '--funcplotuse=', '--pythonfunctions=', '--fofx=sin(-10*x)', '--fponum=true', '--fpofx=x', '--remove=true', '--drawaxis=true']
So looks pretty good. I asked before if it was possible to put the tempfile at the end of the arg list. Isn't that standard convention? I think it will make it easier for any extensions that don't have robust argv parsers. Also the first time I tried it I got a crash. oops.
Aaron
On Fri, 2008-02-29 at 16:13 -0600, Aaron Spike wrote:
Ted Gould wrote:
Aaron, can you get the list of parameters passed again? I think they should be correct now...
So looks pretty good. I asked before if it was possible to put the tempfile at the end of the arg list. Isn't that standard convention? I think it will make it easier for any extensions that don't have robust argv parsers. Also the first time I tried it I got a crash. oops.
I think the file should be at the end now.
What crashed?
--Ted
Ted Gould wrote:
I think the file should be at the end now.
Thanks!
What crashed?
Not sure. :-( It was just a quick test at work and I didn't run in gdb. Seems to be something with the live preview, maybe?
I think I figured out what I find annoying about the live preview. The fact that it reruns during text box edits. Drop down lists and checkboxes make sense to force a rerun. But text boxes seem to be a bit clumsy. Sometimes a text box would work fine. If I'm typing a word to display I could see it regenerate an effect for each added character. But if the effect requires 5 characters and I hit backspace to change the final character, it runs and complains. Likewise in function plotter, if a function starts with "x" and I add "2" in front on the way to typing "2+x" I get a syntax error. Perhaps there should be a tag in the inx files that allows effect authors to specify textboxes where syntax matters and prevent rerunning the effect on every character change, maybe just on lost focus and enter. Thoughts?
Aaron Spike
On Fri, 2008-02-29 at 18:13 -0600, Aaron Spike wrote:
I think I figured out what I find annoying about the live preview. The fact that it reruns during text box edits. Drop down lists and checkboxes make sense to force a rerun. But text boxes seem to be a bit clumsy. Sometimes a text box would work fine. If I'm typing a word to display I could see it regenerate an effect for each added character. But if the effect requires 5 characters and I hit backspace to change the final character, it runs and complains. Likewise in function plotter, if a function starts with "x" and I add "2" in front on the way to typing "2+x" I get a syntax error. Perhaps there should be a tag in the inx files that allows effect authors to specify textboxes where syntax matters and prevent rerunning the effect on every character change, maybe just on lost focus and enter. Thoughts?
Sounds reasonable to me. Kinda like "autocomplete='false'" in HTML. What do you think it should be called? The code is easy, we can just block the passing of the signal.
--Ted
Running win32 nightly build Inkscape0802291457.7z, it looks like both the Measure Path and the Function Plotter are working perfectly, thanks a lot !! They remember the parameter values and the parsing looks right. There is something a bit unusual, namely I can run them once, not twice, second time I seem to get no response. However, that is a separate issue, this is the first time I've ever seen them work, so I'm really impressed.
-Alvin
-----Original Message----- From: inkscape-devel-bounces@lists.sourceforge.net [mailto:inkscape-devel-bounces@lists.sourceforge.net] On Behalf Of Alvin Penner Sent: zaterdag 1 maart 2008 4:27 To: Aaron Spike; Ted Gould; Inkscape Devel List Subject: Re: [Inkscape-devel] Extensions on win32
Running win32 nightly build Inkscape0802291457.7z, it looks like both the Measure Path and the Function Plotter are working perfectly, thanks a lot !!
I think that was with my fix.
Now there should be a somewhat cleaner/better fix, however these two effects are broken again. The effects run successfully, but the parameters are not passed correctly.
-johan
On Sun, Mar 2, 2008 at 9:59 AM, <J.B.C.Engelen@...1578...> wrote:
Now there should be a somewhat cleaner/better fix, however these two effects are broken again. The effects run successfully, but the parameters are not passed correctly.
In rev 17608, function plotter works but dimensions and add nodes do not. Ted, can you please look into whether this might be caused by your recent fixes? And btw please don't forget to copy all your fixes to the 0.46 branch.
On Sun, 2008-03-02 at 20:01 -0400, bulia byak wrote:
On Sun, Mar 2, 2008 at 9:59 AM, <J.B.C.Engelen@...1578...> wrote:
Now there should be a somewhat cleaner/better fix, however these two effects are broken again. The effects run successfully, but the parameters are not passed correctly.
In rev 17608, function plotter works but dimensions and add nodes do not. Ted, can you please look into whether this might be caused by your recent fixes?
With "Add Nodes" I get:
Traceback (most recent call last): File "/home/ted/Development/inkscape.work/install/share/inkscape/extensions/addnodes.py", line 91, in <module> e.affect() File "/home/ted/Development/inkscape.work/install/share/inkscape/extensions/inkex.py", line 154, in affect self.effect() File "/home/ted/Development/inkscape.work/install/share/inkscape/extensions/addnodes.py", line 81, in effect splits = math.ceil(length/self.options.max) ZeroDivisionError: float division
and with Dimensions I get:
/home/ted/Development/inkscape.work/install/share/inkscape/extensions/dimension.py:60: TagNameWarning: Tag names must not contain ':', lxml 2.0 will enforce well-formed tag names as required by the XML specification. marker.set('inkscape:stockid', name)
Aaron, do you have any ideas on what these could be caused by?
And btw please don't forget to copy all your fixes to the 0.46 branch.
I'm not sure what will happen there, there was concern that they contain too much refactoring. I'm not really interesting in fixing the old code as it is really broken (architecturally and code-wise).
--Ted
On Mon, Mar 3, 2008 at 1:00 AM, Ted Gould <ted@...11...> wrote:
I'm not sure what will happen there, there was concern that they contain too much refactoring. I'm not really interesting in fixing the old code as it is really broken (architecturally and code-wise).
I agree that we should push all the extension refactoring to 0.46, but as you see, it all needs testing. I see no other choice but for us to spend at least a few more days testing the extension system after your changes and carefully copying all the fixes over to 0.46.
-----Original Message----- From: bulia byak [mailto:buliabyak@...400...] Sent: maandag 3 maart 2008 1:01 To: Engelen, J.B.C. (Johan) Cc: ted@...11...; inkscape-devel@lists.sourceforge.net; penner@...1856... Subject: Re: [Inkscape-devel] Extensions on win32
On Sun, Mar 2, 2008 at 9:59 AM, <J.B.C.Engelen@...1578...> wrote:
Now there should be a somewhat cleaner/better fix, however
these two effects are broken again. The effects run successfully, but the parameters are not passed correctly.
In rev 17608, function plotter works but dimensions and add nodes do not.
Just checked again. In trunk svn17618, functionplotter is not passed the correct parameters. In 0.46 svn17618, functionplotter *does* work completely (0.46 has only my fix, not Ted's which addresses other issues aswell).
Sorry for the confusion,
Johan
On Mon, 2008-03-03 at 10:37 +0100, J.B.C.Engelen@...1578... wrote:
In rev 17608, function plotter works but dimensions and add nodes do not.
Just checked again. In trunk svn17618, functionplotter is not passed the correct parameters. In 0.46 svn17618, functionplotter *does* work completely (0.46 has only my fix, not Ted's which addresses other issues aswell).
Okay, I am confused. I thought Aaron showed the right parameters with the patch.
So, we now have two branches. On win32 in the RELEASE_0_46_BRANCH branch the parameters to function plotter are correct? But, in trunk they are not?
--Ted
-----Original Message----- From: Ted Gould [mailto:ted@...11...] Sent: maandag 3 maart 2008 19:32 To: Engelen, J.B.C. (Johan) Cc: buliabyak@...400...; inkscape-devel@lists.sourceforge.net; penner@...1856... Subject: RE: [Inkscape-devel] Extensions on win32
On Mon, 2008-03-03 at 10:37 +0100, J.B.C.Engelen@...1578... wrote:
In rev 17608, function plotter works but dimensions and
add nodes do
not.
Just checked again. In trunk svn17618, functionplotter is
not passed
the correct parameters. In 0.46 svn17618, functionplotter
*does* work
completely (0.46 has only my fix, not Ted's which addresses other issues aswell).
Okay, I am confused. I thought Aaron showed the right parameters with the patch.
So, we now have two branches. On win32 in the RELEASE_0_46_BRANCH branch the parameters to function plotter are correct? But, in trunk they are not?
--Ted
Correct.
On Mon, 2008-03-03 at 12:48 -0600, Aaron Spike wrote:
Ted Gould wrote:
Okay, I am confused. I thought Aaron showed the right parameters with the patch.
My test was on trunk.
So, we now have two branches. On win32 in the RELEASE_0_46_BRANCH branch the parameters to function plotter are correct? But, in trunk they are not?
Aaron, is your test something I could run? I'm just all confused, I think that's the best source of data, I'd like to get runs in various setups to try and figure out what's happening.
Thanks, Ted
Ted Gould wrote:
Aaron, is your test something I could run? I'm just all confused, I think that's the best source of data, I'd like to get runs in various setups to try and figure out what's happening.
I've just been adding:
debug(sys.argv)
To the affect() method in inkex.py (around line 149 or so). The debug procedure is defined near the beginning of the file. Simply wraps its argument in str() and writes to sys.stderr. For some of the problems we're seeing on Vista it may also be interesting to view sys.path.
Aaron Spike
participants (8)
-
unknown＠example.com -
Aaron Spike -
Alvin Penner -
Bob Jamison -
bulia byak -
MenTaLguY -
Simon Mieth -
Ted Gould