[GRASS-dev] r57990 breaks r.colors colour selector

GRASS GIS GRASS GIS developer
1

Hi Martin,
with r57990 on my system after selecting colour from the list, it
doesn't get appended to executed command and thus it's not possible to
set predefined colour with gui.
Probably changes in some other places are needed too.

Python: 2.7.5
wxPython: 2.9.4.1
Maris.

2

Hi,

2013/10/15 Maris Nartiss <maris.gis@gmail.com>:

with r57990 on my system after selecting colour from the list, it
doesn't get appended to executed command and thus it's not possible to
set predefined colour with gui.
Probably changes in some other places are needed too.

I cannot reproduce this error, in other words, it's working for me.

Python: 2.7.5
wxPython: 2.9.4.1

Python: 2.7.5
wxPython: 2.8.12.1

Is there any wxGUI developer who has wxPython 2.9 installed on her/his machine?

Martin

--
Martin Landa <landa.martin gmail.com> * http://geo.fsv.cvut.cz/~landa

3

Hi,

2013/10/15 Martin Landa <landa.martin@gmail.com>:

with r57990 on my system after selecting colour from the list, it
doesn't get appended to executed command and thus it's not possible to
set predefined colour with gui.
Probably changes in some other places are needed too.

I cannot reproduce this error, in other words, it's working for me.

please try the attached patch. Unfortunately there is a side-effect,
when deleting text from the widget I get well-known

(python:21174): Pango-CRITICAL **: pango_layout_get_cursor_pos:
assertion `index >= 0 && index <= layout->length' failed

Martin

--
Martin Landa <landa.martin gmail.com> * http://geo.fsv.cvut.cz/~landa

(attachments)

color.diff (769 Bytes)

4

The issue is that currently GRASS wxgui is not enforcing wxpython
version. On my system I have both - 2.8 and 2.9. As no specific
version is requested, python is loading one of them - in my case it's
2.9 (my guess as 2.9>2.8). At least on my system is present wxversion,
that allows to select specific wxpython version on load [1].

Still on other systems wxversion might be missing.

1. http://www.wxpython.org/docs/api/wxversion-module.html

2013/10/15 Martin Landa <landa.martin@gmail.com>:

Hi,

2013/10/15 Maris Nartiss <maris.gis@gmail.com>:

with r57990 on my system after selecting colour from the list, it
doesn't get appended to executed command and thus it's not possible to
set predefined colour with gui.
Probably changes in some other places are needed too.

I cannot reproduce this error, in other words, it's working for me.

Python: 2.7.5
wxPython: 2.9.4.1

Python: 2.7.5
wxPython: 2.8.12.1

Is there any wxGUI developer who has wxPython 2.9 installed on her/his machine?

Martin

--
Martin Landa <landa.martin gmail.com> * http://geo.fsv.cvut.cz/~landa

5

Ignoring the common side effect, it works just fine.
Commit it, please.

Thanks,
Maris.

2013/10/15 Martin Landa <landa.martin@gmail.com>:

Hi,

2013/10/15 Martin Landa <landa.martin@gmail.com>:

with r57990 on my system after selecting colour from the list, it
doesn't get appended to executed command and thus it's not possible to
set predefined colour with gui.
Probably changes in some other places are needed too.

I cannot reproduce this error, in other words, it's working for me.

please try the attached patch. Unfortunately there is a side-effect,
when deleting text from the widget I get well-known

(python:21174): Pango-CRITICAL **: pango_layout_get_cursor_pos:
assertion `index >= 0 && index <= layout->length' failed

Martin

--
Martin Landa <landa.martin gmail.com> * http://geo.fsv.cvut.cz/~landa

6

Hi,

2013/10/15 Maris Nartiss <maris.gis@gmail.com>:

The issue is that currently GRASS wxgui is not enforcing wxpython
version. On my system I have both - 2.8 and 2.9. As no specific
version is requested, python is loading one of them - in my case it's
2.9 (my guess as 2.9>2.8). At least on my system is present wxversion,
that allows to select specific wxpython version on load [1].

well, but it will not solve the reported bug. When wxPython 3.0 will
be released and used across major GNU/Linux distributions, the bug
will appear again. Martin

7

Contrary. It's the right way to go. When 3.0 will be released, most
likely it will live for some time in parallel with 2.8 thus we can
request 2.8 even when 3.0 is present. Besides after 3.0 release some
users will have 3.1 installed and the story repeats.
I can't write code now thus I can only suggest to enforce 2.8 and
then, when we are ready, bump to 3.0. It should be made in a way to
easy enable 2.9 enforcement for testing.

Besides - my bug reports about issues with 2.9 were rejected as "it's
a development version".

Just my 0.2
Thanks again, Martin.

Maris.

2013/10/15 Martin Landa <landa.martin@gmail.com>:

Hi,

2013/10/15 Maris Nartiss <maris.gis@gmail.com>:

The issue is that currently GRASS wxgui is not enforcing wxpython
version. On my system I have both - 2.8 and 2.9. As no specific
version is requested, python is loading one of them - in my case it's
2.9 (my guess as 2.9>2.8). At least on my system is present wxversion,
that allows to select specific wxpython version on load [1].

well, but it will not solve the reported bug. When wxPython 3.0 will
be released and used across major GNU/Linux distributions, the bug
will appear again. Martin

8

Hi,

2013/10/15 Maris Nartiss <maris.gis@gmail.com>:

Ignoring the common side effect, it works just fine.
Commit it, please.

please could you try a new version of the patch which apparently
doesn't produce Pango message?

Martin

(attachments)

color-1.diff (934 Bytes)

9

Yep, this one works just fine and there are no error messages.

Still I would *love* to see an easy way how to switch to stable
wxpython as currently with 2.9 vector import dialog, profile tool,
category editing etc. are broken not to mention constant Yield():
wxYield called recursively and similar error messages. Seems that
there will be quite a work to prepare for 3.0.

Maris.

2013/10/15 Martin Landa <landa.martin@gmail.com>:

Hi,

2013/10/15 Maris Nartiss <maris.gis@gmail.com>:

Ignoring the common side effect, it works just fine.
Commit it, please.

please could you try a new version of the patch which apparently
doesn't produce Pango message?

Martin

10

Hi,

2013/10/16 Maris Nartiss <maris.gis@gmail.com>:

Yep, this one works just fine and there are no error messages.

committed as r58012. Martin

--
Martin Landa <landa.martin gmail.com> * http://geo.fsv.cvut.cz/~landa