[GRASS-dev] Windows native GRASS build

Happy new year all!

For those of you following the GRASS on windows efforts with QGIS, I
have now got GRASS building with all but about 20 modules under
msys/windows. I havent compiled in postgresql support yet.. GRASS
attribute viewing and man page viewing are now working under windows
qgis for me. There are two small issues remaining that I noticed (and
maybe more that others will notice):

1) Every time a layer is added a black console box temporarily appears
and then disappears.
2) When clicking on a tool inthe GRASS toolbox, I get a message like:
   "Warning: cannot find key output"
3) Are any of the unbuilt modules listed below particularly vital to
have in the QGIS 0.8 binary installer for windows?

I will upload a binary for testing tonight or tomorrow.

I compiled with the following options:

./configure --build=i686-pc-mingw32 \
            --with-jpeg-libs=/bin \
            --with-libs=/local/lib/ \
            --with-png-libs=/bin \
            --with-fftw-libs=/local/lib \
            --with-fftw-includes=/local/include \
            --with-includes=/local/include \
            --with-includes=/include \
            --with-tcltk=no \
            --with-postgres=no \
            --with-opengl=no

And the end of make report looks like this:

----------------------------------------------------------------------
GRASS GIS compilation log
-------------------------
Started compilation: Sun Dec 31 20:15:24 CST 2006
--
Errors in:
/src/grass6/display/drivers/PNG
/src/grass6/imagery/i.class

/src/grass6/imagery/i.ortho.photo/photo.2image
/src/grass6/imagery/i.ortho.photo/photo.2target
/src/grass6/imagery/i.points
/src/grass6/imagery/i.vpoints
/src/grass6/raster/r.li/r.li.daemon
/src/grass6/raster/r.li/r.li.edgedensity
/src/grass6/raster/r.li/r.li.patchdensity
/src/grass6/raster/r.li/r.li.patchnumber
/src/grass6/raster/r.li/r.li.shape
/src/grass6/raster/r.li/r.li.simpson
/src/grass6/raster/r.li/r.li.shannon
/src/grass6/raster/r.li/r.li.meanPatchSize
/src/grass6/raster/r.li/r.li.meanPixelAttribute
/src/grass6/raster/r.li/r.li.patchAreaDistributionCV
/src/grass6/raster/r.li/r.li.patchAreaDistributionSD
/src/grass6/raster/r.li/r.li.patchAreaDistributionRANGE
/src/grass6/raster/r.li/r.li.contrastWeightedEdgeDensity
/src/grass6/raster/r.li/r.li.richness
/src/grass6/raster/r.li/r.li.dominance
--
Finished compilation: Sun Dec 31 20:47:48 CST 2006
(In case of errors please change into the directory with error and run 'make')
make: *** [default] Error 1

--
Tim Sutton

Visit http://qgis.org for a great Open Source GIS
Home Page: http://linfiniti.com
Skype: timlinux
MSN: tim_bdworld@msn.com
Yahoo: tim_bdworld@yahoo.com
Jabber: timlinux
Irc: timlinux on QGIS at freenode.net

On 1/1/07, Tim Sutton <tim@linfiniti.com> wrote:

Happy new year all!

For those of you following the GRASS on windows efforts with QGIS, I
have now got GRASS building with all but about 20 modules under
msys/windows. I havent compiled in postgresql support yet.. GRASS
attribute viewing and man page viewing are now working under windows
qgis for me. There are two small issues remaining that I noticed (and
maybe more that others will notice):

1) Every time a layer is added a black console box temporarily appears
and then disappears.

It is known problem, db driver is exe and when it is started by spawnl()
it opens Windows console for a while. I am not Win expert and I dont know
how to avoid this. I have asked already in QGIS and GRASS devel lists.

2) When clicking on a tool inthe GRASS toolbox, I get a message like:
   "Warning: cannot find key output"

Which module?

3) Are any of the unbuilt modules listed below particularly vital to
have in the QGIS 0.8 binary installer for windows?

I will upload a binary for testing tonight or tomorrow.

I compiled with the following options:

./configure --build=i686-pc-mingw32 \
            --with-jpeg-libs=/bin \
            --with-libs=/local/lib/ \
            --with-png-libs=/bin \
            --with-fftw-libs=/local/lib \
            --with-fftw-includes=/local/include \
            --with-includes=/local/include \
            --with-includes=/include \
            --with-tcltk=no \
            --with-postgres=no \

Postgres is important IMO.

            --with-opengl=no

And the end of make report looks like this:

----------------------------------------------------------------------
GRASS GIS compilation log
-------------------------
Started compilation: Sun Dec 31 20:15:24 CST 2006
--
Errors in:
/src/grass6/display/drivers/PNG

Not used.

/src/grass6/imagery/i.class
/src/grass6/imagery/i.ortho.photo/photo.2image
/src/grass6/imagery/i.ortho.photo/photo.2target
/src/grass6/imagery/i.points
/src/grass6/imagery/i.vpoints

Some i.* modules which depend on GRASS monitor must be
impemented in QGIS in future. No solution for now.

/src/grass6/raster/r.li/r.li.daemon
/src/grass6/raster/r.li/r.li.edgedensity
/src/grass6/raster/r.li/r.li.patchdensity
/src/grass6/raster/r.li/r.li.patchnumber
/src/grass6/raster/r.li/r.li.shape
/src/grass6/raster/r.li/r.li.simpson
/src/grass6/raster/r.li/r.li.shannon
/src/grass6/raster/r.li/r.li.meanPatchSize
/src/grass6/raster/r.li/r.li.meanPixelAttribute
/src/grass6/raster/r.li/r.li.patchAreaDistributionCV
/src/grass6/raster/r.li/r.li.patchAreaDistributionSD
/src/grass6/raster/r.li/r.li.patchAreaDistributionRANGE
/src/grass6/raster/r.li/r.li.contrastWeightedEdgeDensity
/src/grass6/raster/r.li/r.li.richness
/src/grass6/raster/r.li/r.li.dominance

Not crucial.

Radim

Finished compilation: Sun Dec 31 20:47:48 CST 2006
(In case of errors please change into the directory with error and run 'make')
make: *** [default] Error 1

--
Tim Sutton

Visit http://qgis.org for a great Open Source GIS
Home Page: http://linfiniti.com
Skype: timlinux
MSN: tim_bdworld@msn.com
Yahoo: tim_bdworld@yahoo.com
Jabber: timlinux
Irc: timlinux on QGIS at freenode.net
_______________________________________________
Qgis-developer mailing list
Qgis-developer@lists.qgis.org
http://lists.qgis.org/cgi-bin/mailman/listinfo/qgis-developer

Hi

> 1) Every time a layer is added a black console box temporarily appears
> and then disappears.

It is known problem, db driver is exe and when it is started by spawnl()
it opens Windows console for a while. I am not Win expert and I dont know
how to avoid this. I have asked already in QGIS and GRASS devel lists.

Ok - I think its not a big problem for now.

> 2) When clicking on a tool inthe GRASS toolbox, I get a message like:
> "Warning: cannot find key output"

Which module?

*every module*

Any ideas what can be causing that?

> 3) Are any of the unbuilt modules listed below particularly vital to
> have in the QGIS 0.8 binary installer for windows?

Postgres is important IMO.

Yes I will work tonight on getting pg support compiled under msys.

Not crucial.

Ok I will ignore them for this release exercise.

Regards

Tim

Radim

> Finished compilation: Sun Dec 31 20:47:48 CST 2006
> (In case of errors please change into the directory with error and run 'make')
> make: *** [default] Error 1
>
> --
> Tim Sutton
>
> Visit http://qgis.org for a great Open Source GIS
> Home Page: http://linfiniti.com
> Skype: timlinux
> MSN: tim_bdworld@msn.com
> Yahoo: tim_bdworld@yahoo.com
> Jabber: timlinux
> Irc: timlinux on QGIS at freenode.net
> _______________________________________________
> Qgis-developer mailing list
> Qgis-developer@lists.qgis.org
> http://lists.qgis.org/cgi-bin/mailman/listinfo/qgis-developer
>

--
Tim Sutton

Visit http://qgis.org for a great Open Source GIS
Home Page: http://linfiniti.com
Skype: timlinux
MSN: tim_bdworld@msn.com
Yahoo: tim_bdworld@yahoo.com
Jabber: timlinux
Irc: timlinux on QGIS at freenode.net

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Radim Blazek ha scritto:

On 1/1/07, Tim Sutton <tim@linfiniti.com> wrote:

./configure --build=i686-pc-mingw32 \
            --with-jpeg-libs=/bin \
            --with-libs=/local/lib/ \
            --with-png-libs=/bin \
            --with-fftw-libs=/local/lib \
            --with-fftw-includes=/local/include \
            --with-includes=/local/include \
            --with-includes=/include \
            --with-tcltk=no \
            --with-postgres=no \

Postgres is important IMO.

Very important for any serious use of grass.

/src/grass6/raster/r.li/r.li.daemon
/src/grass6/raster/r.li/r.li.edgedensity
/src/grass6/raster/r.li/r.li.patchdensity
/src/grass6/raster/r.li/r.li.patchnumber
/src/grass6/raster/r.li/r.li.shape
/src/grass6/raster/r.li/r.li.simpson
/src/grass6/raster/r.li/r.li.shannon
/src/grass6/raster/r.li/r.li.meanPatchSize
/src/grass6/raster/r.li/r.li.meanPixelAttribute
/src/grass6/raster/r.li/r.li.patchAreaDistributionCV
/src/grass6/raster/r.li/r.li.patchAreaDistributionSD
/src/grass6/raster/r.li/r.li.patchAreaDistributionRANGE
/src/grass6/raster/r.li/r.li.contrastWeightedEdgeDensity
/src/grass6/raster/r.li/r.li.richness
/src/grass6/raster/r.li/r.li.dominance

Not crucial.

Agreed, but we would like to have it. Tim, could you please forward the
errors to Serena Pallecchi, Davide Spano and Claudio Porta?
Thanks.
pc
- --
Paolo Cavallini
email+jabber: cavallini@faunalia.it
www.faunalia.it
Piazza Garibaldi 5 - 56025 Pontedera (PI), Italy Tel: (+39)348-3801953
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFmgnt/NedwLUzIr4RAsAeAJ9wK/2CTtTv49B0EyJDx5N/CJHxngCcCcyI
5czwr8c64Luk4CvCXbwQOp4=
=t+rs
-----END PGP SIGNATURE-----

Hello Paolo

On Tue, 2 Jan 2007, Paolo Cavallini wrote:

[...]

/src/grass6/raster/r.li/r.li.daemon
/src/grass6/raster/r.li/r.li.edgedensity
/src/grass6/raster/r.li/r.li.patchdensity
/src/grass6/raster/r.li/r.li.patchnumber
/src/grass6/raster/r.li/r.li.shape
/src/grass6/raster/r.li/r.li.simpson
/src/grass6/raster/r.li/r.li.shannon
/src/grass6/raster/r.li/r.li.meanPatchSize
/src/grass6/raster/r.li/r.li.meanPixelAttribute
/src/grass6/raster/r.li/r.li.patchAreaDistributionCV
/src/grass6/raster/r.li/r.li.patchAreaDistributionSD
/src/grass6/raster/r.li/r.li.patchAreaDistributionRANGE
/src/grass6/raster/r.li/r.li.contrastWeightedEdgeDensity
/src/grass6/raster/r.li/r.li.richness
/src/grass6/raster/r.li/r.li.dominance

Not crucial.

Agreed, but we would like to have it. Tim, could you please forward the
errors to Serena Pallecchi, Davide Spano and Claudio Porta?

Please see attached. Main problems as far as I can see:
- Use of FIFOs? AIUI Windows doesn't have FIFOs so that won't work.
- Also I don't think fork() is available on Windows either - Glynn may be able to advise.

Paul

(attachments)

make.log (17.6 KB)

On 1/1/07, Tim Sutton <tim@linfiniti.com> wrote:

> > 2) When clicking on a tool inthe GRASS toolbox, I get a message like:
> > "Warning: cannot find key output"
>
> Which module?

*every module*

Any ideas what can be causing that?

I cannot get that message with
http://qgis.org/uploadfiles/testbuilds/qgis_setup.exe
(02-Jan-2007 06:29 44M).

No idea, very strange. It was always working. The message means that
parameter 'output' is missing in the XML description of GRASS module
written out by a module run with --interface-description option.
You could try to run a module (in shell) with that option and check if
the 'output' option tag is in XML printed on stdout.

Radim

Hi

Yes thats a new build you tested with. I uploaded it I went to sleep
at 4am this morning. I was going to write a message when I woke up
saying the problem was fixed but looks like you beat me to it ! :stuck_out_tongue:
The version you have downloaded also has GRASS built with postgres
support. I was not able to test pg support at home - can someone give
it a try?

I'm hoping the version I uploaded last night

http://qgis.org/uploadfiles/testbuilds/qgis_setup.exe

fixes all remaining issues so everyone interested in the Win build
please download it and test it out. Please note that the installer
offers to download sample data for you but you will need to manually
unzip the data in you Quantum GIS/SampleData dir since I couldnt yet
get NSIS to automate that.

I hope to put this version online on the main qgis 0.8 release page
later today so any issues found after that will have to get resolved
in a follow up bug fix release. I will also upload the msys
environment which contains everything you need to build grass for
windows msys later today.

Regards

Tim

On 1/2/07, Radim Blazek <radim.blazek@gmail.com> wrote:

On 1/1/07, Tim Sutton <tim@linfiniti.com> wrote:
> > > 2) When clicking on a tool inthe GRASS toolbox, I get a message like:
> > > "Warning: cannot find key output"
> >
> > Which module?
>
> *every module*
>
> Any ideas what can be causing that?

I cannot get that message with
http://qgis.org/uploadfiles/testbuilds/qgis_setup.exe
(02-Jan-2007 06:29 44M).

No idea, very strange. It was always working. The message means that
parameter 'output' is missing in the XML description of GRASS module
written out by a module run with --interface-description option.
You could try to run a module (in shell) with that option and check if
the 'output' option tag is in XML printed on stdout.

Radim

--
Tim Sutton

Visit http://qgis.org for a great Open Source GIS
Home Page: http://linfiniti.com
Skype: timlinux
MSN: tim_bdworld@msn.com
Yahoo: tim_bdworld@yahoo.com
Jabber: timlinux
Irc: timlinux on QGIS at freenode.net

On Tue, 2007-01-02 at 11:33 +0000, Paul Kelly wrote:

Hello Paolo

On Tue, 2 Jan 2007, Paolo Cavallini wrote:

[...]
>>> /src/grass6/raster/r.li/r.li.daemon
>>> /src/grass6/raster/r.li/r.li.edgedensity
>>> /src/grass6/raster/r.li/r.li.patchdensity
>>> /src/grass6/raster/r.li/r.li.patchnumber
>>> /src/grass6/raster/r.li/r.li.shape
>>> /src/grass6/raster/r.li/r.li.simpson
>>> /src/grass6/raster/r.li/r.li.shannon
>>> /src/grass6/raster/r.li/r.li.meanPatchSize
>>> /src/grass6/raster/r.li/r.li.meanPixelAttribute
>>> /src/grass6/raster/r.li/r.li.patchAreaDistributionCV
>>> /src/grass6/raster/r.li/r.li.patchAreaDistributionSD
>>> /src/grass6/raster/r.li/r.li.patchAreaDistributionRANGE
>>> /src/grass6/raster/r.li/r.li.contrastWeightedEdgeDensity
>>> /src/grass6/raster/r.li/r.li.richness
>>> /src/grass6/raster/r.li/r.li.dominance
>>
>> Not crucial.
>
> Agreed, but we would like to have it. Tim, could you please forward the
> errors to Serena Pallecchi, Davide Spano and Claudio Porta?

Please see attached. Main problems as far as I can see:
- Use of FIFOs? AIUI Windows doesn't have FIFOs so that won't work.
- Also I don't think fork() is available on Windows either - Glynn may be
able to advise.

Linux: fork() exec()
Win32: CreateProcess()

Linux: exit()
Win32: ExitProcess()

Linux: get/putenv()
Win32: Get/SetEnvironmentVariable()

Linux: waitpid()
Win32: GetExitCodeProcess()

Linux; getpid()
Win32: GetCurrentProcessId()

It would be nice to have wrappers for these functions. Thoughts, Glynn?

--
Brad Douglas <rez touchofmadness com> KB8UYR/6
Address: 37.493,-121.924 / WGS84 National Map Corps #TNMC-3785

Hi

For those interested, I have uploaded the complete msys environemnt
that Tisham Dhar and myself set up here:

http://qgis.org/uploadfiles/msys/msys.tar.gz

Its around 230mb since I have not run make clean in the source dirs
and stripped away unneeeded stuff. However it contains a complete
build environment for making grass (along with numerous other FOSS GIS
stuff) under msys / windows. The GRASS in there is an anonynous
checkout from head and a cvs client is included in teh msys so you can
cvs update grass from in the msys shell.

Regards

Tim

On 1/2/07, Brad Douglas <rez@touchofmadness.com> wrote:

On Tue, 2007-01-02 at 11:33 +0000, Paul Kelly wrote:
> Hello Paolo
>
> On Tue, 2 Jan 2007, Paolo Cavallini wrote:
>
> [...]
> >>> /src/grass6/raster/r.li/r.li.daemon
> >>> /src/grass6/raster/r.li/r.li.edgedensity
> >>> /src/grass6/raster/r.li/r.li.patchdensity
> >>> /src/grass6/raster/r.li/r.li.patchnumber
> >>> /src/grass6/raster/r.li/r.li.shape
> >>> /src/grass6/raster/r.li/r.li.simpson
> >>> /src/grass6/raster/r.li/r.li.shannon
> >>> /src/grass6/raster/r.li/r.li.meanPatchSize
> >>> /src/grass6/raster/r.li/r.li.meanPixelAttribute
> >>> /src/grass6/raster/r.li/r.li.patchAreaDistributionCV
> >>> /src/grass6/raster/r.li/r.li.patchAreaDistributionSD
> >>> /src/grass6/raster/r.li/r.li.patchAreaDistributionRANGE
> >>> /src/grass6/raster/r.li/r.li.contrastWeightedEdgeDensity
> >>> /src/grass6/raster/r.li/r.li.richness
> >>> /src/grass6/raster/r.li/r.li.dominance
> >>
> >> Not crucial.
> >
> > Agreed, but we would like to have it. Tim, could you please forward the
> > errors to Serena Pallecchi, Davide Spano and Claudio Porta?
>
> Please see attached. Main problems as far as I can see:
> - Use of FIFOs? AIUI Windows doesn't have FIFOs so that won't work.
> - Also I don't think fork() is available on Windows either - Glynn may be
> able to advise.

Linux: fork() exec()
Win32: CreateProcess()

Linux: exit()
Win32: ExitProcess()

Linux: get/putenv()
Win32: Get/SetEnvironmentVariable()

Linux: waitpid()
Win32: GetExitCodeProcess()

Linux; getpid()
Win32: GetCurrentProcessId()

It would be nice to have wrappers for these functions. Thoughts, Glynn?

--
Brad Douglas <rez touchofmadness com> KB8UYR/6
Address: 37.493,-121.924 / WGS84 National Map Corps #TNMC-3785

_______________________________________________
grass-dev mailing list
grass-dev@grass.itc.it
http://grass.itc.it/mailman/listinfo/grass-dev

--
Tim Sutton

Visit http://qgis.org for a great Open Source GIS
Home Page: http://linfiniti.com
Skype: timlinux
MSN: tim_bdworld@msn.com
Yahoo: tim_bdworld@yahoo.com
Jabber: timlinux
Irc: timlinux on QGIS at freenode.net

Thanks Marco!
-mwindows helped, attached is patch for Grass.make.in
and dbf driver Makefile. Similar change must be done in
Makefiles of all drivers. configure must be run because
Grass.make.in is source for Grass.make

It would be nice if this patch or something similar could be applied
to GRASS CVS head and 6.2 release branch.

Tim can you patch your GRASS in MSYS and create a new build?
We dont have to wait for 0.8.1 IMO because it does not include
changes in QGIS, it is just a new binary build.

Radim

On 1/3/07, Marco Hugentobler <marco.hugentobler@karto.baug.ethz.ch> wrote:

Hi Radim and Tim,

If the db driver is compiled with mingw gcc and you have the possibility to
recompile it yourself, you may use the linker option -mwindows. I don't know
of other possibilities to avoid the console (but I am no win expert too).

Marco

Am Montag, 1. Januar 2007 18:30 schrieb Tim Sutton:
> Hi
>
> > > 1) Every time a layer is added a black console box temporarily appears
> > > and then disappears.
> >
> > It is known problem, db driver is exe and when it is started by spawnl()
> > it opens Windows console for a while. I am not Win expert and I dont know
> > how to avoid this. I have asked already in QGIS and GRASS devel lists.
>
> Ok - I think its not a big problem for now.
>
_______________________________________________
Qgis-developer mailing list
Qgis-developer@lists.qgis.org
http://lists.qgis.org/cgi-bin/mailman/listinfo/qgis-developer

(attachments)

Grass.make.in.patch (364 Bytes)
Makefile.patch (412 Bytes)

Brad Douglas wrote:

> >>> /src/grass6/raster/r.li/r.li.daemon
> >>> /src/grass6/raster/r.li/r.li.edgedensity
> >>> /src/grass6/raster/r.li/r.li.patchdensity
> >>> /src/grass6/raster/r.li/r.li.patchnumber
> >>> /src/grass6/raster/r.li/r.li.shape
> >>> /src/grass6/raster/r.li/r.li.simpson
> >>> /src/grass6/raster/r.li/r.li.shannon
> >>> /src/grass6/raster/r.li/r.li.meanPatchSize
> >>> /src/grass6/raster/r.li/r.li.meanPixelAttribute
> >>> /src/grass6/raster/r.li/r.li.patchAreaDistributionCV
> >>> /src/grass6/raster/r.li/r.li.patchAreaDistributionSD
> >>> /src/grass6/raster/r.li/r.li.patchAreaDistributionRANGE
> >>> /src/grass6/raster/r.li/r.li.contrastWeightedEdgeDensity
> >>> /src/grass6/raster/r.li/r.li.richness
> >>> /src/grass6/raster/r.li/r.li.dominance
> >>
> >> Not crucial.
> >
> > Agreed, but we would like to have it. Tim, could you please forward the
> > errors to Serena Pallecchi, Davide Spano and Claudio Porta?
>
> Please see attached. Main problems as far as I can see:
> - Use of FIFOs? AIUI Windows doesn't have FIFOs so that won't work.
> - Also I don't think fork() is available on Windows either - Glynn may be
> able to advise.

Linux: fork() exec()
Win32: CreateProcess()

Linux: exit()
Win32: ExitProcess()

Linux: get/putenv()
Win32: Get/SetEnvironmentVariable()

Linux: waitpid()
Win32: GetExitCodeProcess()

Linux; getpid()
Win32: GetCurrentProcessId()

It would be nice to have wrappers for these functions. Thoughts, Glynn?

There are limits to "wrapping" process management. It isn't as if Unix
and Windows are the same except for the names given to the functions;
the semantics are quite different. Particularly the Unix model of
separate fork() and exec(), with the ability to substantially alter
the state of the child between the two, versus a single
CreateProcess() call with a fixed set of parameters.

Having said that, you could wrap the simpler cases. You would need to
stick to what's available on Windows; some of the more complex things
which can be done on Unix either aren't possible on Windows (e.g.
setting descriptors other than stdin/stdout/stderr) or don't have any
equivalent there (e.g. signal handling).

I already started writing potential replacements for typical usage of
system(); see lib/gis/spawn.c. G_spawn() is a direct replacement for
the simplest uses of system(), except that it accepts individual
arguments rather than a single string and doesn't use "sh -c ...".
G_spawn_ex() is intended to handle more advanced usage, e.g.
redirection, signal handling, environment variable substitution and
binding.

--
Glynn Clements <glynn@gclements.plus.com>

Hi Radim

I can make another build but probably not until monday. If the patches
go into grass cvs that will be great cause then I can just update &
build.

Regards

Tim

On 1/5/07, Radim Blazek <radim.blazek@gmail.com> wrote:

Thanks Marco!
-mwindows helped, attached is patch for Grass.make.in
and dbf driver Makefile. Similar change must be done in
Makefiles of all drivers. configure must be run because
Grass.make.in is source for Grass.make

It would be nice if this patch or something similar could be applied
to GRASS CVS head and 6.2 release branch.

Tim can you patch your GRASS in MSYS and create a new build?
We dont have to wait for 0.8.1 IMO because it does not include
changes in QGIS, it is just a new binary build.

Radim

On 1/3/07, Marco Hugentobler <marco.hugentobler@karto.baug.ethz.ch> wrote:
>
> Hi Radim and Tim,
>
> If the db driver is compiled with mingw gcc and you have the possibility to
> recompile it yourself, you may use the linker option -mwindows. I don't know
> of other possibilities to avoid the console (but I am no win expert too).
>
> Marco
>
> Am Montag, 1. Januar 2007 18:30 schrieb Tim Sutton:
> > Hi
> >
> > > > 1) Every time a layer is added a black console box temporarily appears
> > > > and then disappears.
> > >
> > > It is known problem, db driver is exe and when it is started by spawnl()
> > > it opens Windows console for a while. I am not Win expert and I dont know
> > > how to avoid this. I have asked already in QGIS and GRASS devel lists.
> >
> > Ok - I think its not a big problem for now.
> >
> _______________________________________________
> Qgis-developer mailing list
> Qgis-developer@lists.qgis.org
> http://lists.qgis.org/cgi-bin/mailman/listinfo/qgis-developer
>

--
Tim Sutton

Visit http://qgis.org for a great Open Source GIS
Home Page: http://linfiniti.com
Skype: timlinux
MSN: tim_bdworld@msn.com
Yahoo: tim_bdworld@yahoo.com
Jabber: timlinux
Irc: timlinux on QGIS at freenode.net

On Mon, 1 Jan 2007, Radim Blazek wrote:

On 1/1/07, Tim Sutton <tim@linfiniti.com> wrote:

Happy new year all!

For those of you following the GRASS on windows efforts with QGIS, I
have now got GRASS building with all but about 20 modules under
msys/windows. I havent compiled in postgresql support yet.. GRASS
attribute viewing and man page viewing are now working under windows
qgis for me. There are two small issues remaining that I noticed (and
maybe more that others will notice):

1) Every time a layer is added a black console box temporarily appears
and then disappears.

It is known problem, db driver is exe and when it is started by spawnl()
it opens Windows console for a while. I am not Win expert and I dont know
how to avoid this. I have asked already in QGIS and GRASS devel lists.

Just wondering, if the newly improved G_spawn() might be a possible tidier solution for this? Not sure and not really able to test patches here yet as my Windows build is still having the db driver communication problem described by Moritz in earlier e-mails. Must get round to trying it with a different database.
Just wanted to put it on the list anyway. Sorry can't take the idea any further right now, but the relevant file is lib/db/dbmi_client/start.c

Paul