Unfortunatly GRASS has no Windows developers. Developers use Linux or
Mac and not windows and thus fixing Windows specific bugs is hard.
I have attached patch to ticket where such problem is described. http://trac.osgeo.org/grass/ticket/629
Please, apply patch to GRASS 6.4/5, recompile and run with DEBUG=1.
Post Your results to ticket.
Thanks,
Maris.
2009/9/27, grassyjohn <jf_walsh@hotmail.com>:
I've tried importing the spearfish_contours files via File-->Import Vector
Map-->Multiple Formats Using OGR and get this message
Invalid argument: can't _spawnl
Unable to start driver <sqlite>
Unable open database
<D:/GIS_DataBase/Spearfish60/user1/dbf/spearfish_contours.db> by driver
<sqlite>
I've tried using dbf and ogr database drivers but the result is the same.
Thanks for the tip but I've no idea of how to "re-compile" . Can this be
done from within the software or do I need to use something like Dev C++?
I have a box running on Vector Linux so I may just install on there instead.
Maris Nartiss wrote:
Unfortunatly GRASS has no Windows developers. Developers use Linux or
Mac and not windows and thus fixing Windows specific bugs is hard.
I have attached patch to ticket where such problem is described. http://trac.osgeo.org/grass/ticket/629
Please, apply patch to GRASS 6.4/5, recompile and run with DEBUG=1.
Post Your results to ticket.
Thanks,
Maris.
2009/9/27, grassyjohn <jf_walsh@hotmail.com>:
I've tried importing the spearfish_contours files via File-->Import
Vector
Map-->Multiple Formats Using OGR and get this message
Invalid argument: can't _spawnl
Unable to start driver <sqlite>
Unable open database
<D:/GIS_DataBase/Spearfish60/user1/dbf/spearfish_contours.db> by driver
<sqlite>
I've tried using dbf and ogr database drivers but the result is the same.
I have attached patch to ticket where such problem is
described. http://trac.osgeo.org/grass/ticket/629
Please, apply patch to GRASS 6.4/5, recompile and run with
DEBUG=1.
Post Your results to ticket.
sorry I just noticed that was a brand new patch in the tracker today,
it won't be in last week's binary.
Also previous patch should not be commited to anywhere. I just want to
know (get info) what's going on when db spawn call fails. If somebody
can recompile with that tickets patch and then post results to trac,
we may get an idea what's failing. I suspect forward vs backward
slashes issue. Still without exact info it's hard to tell. Also unless
somebody is going to test various patches/ticket fixes on windows,
windows version will suck hard (die, die!).
Maris (revorking bits in ogsf).
2009/9/28, Hamish <hamish_b@yahoo.com>:
Maris wrote:
I have attached patch to ticket where such problem is
described. http://trac.osgeo.org/grass/ticket/629
Please, apply patch to GRASS 6.4/5, recompile and run with
DEBUG=1.
Post Your results to ticket.
sorry I just noticed that was a brand new patch in the tracker today,
it won't be in last week's binary.
if you use the latest native windows installer the patch should already
be applied. (WinGRASS-6.4.0SVN-r39271-1-Setup.exe 21 Sept 2009)
if you are using that version and it still breaks, we'd like to hear
about it.
I did use the latest installer on XP. If I am importing ESRI files and the
corresponding dbf file, does it matter where I put them as long as I select
the correct paths to the files in the dialogs?
> I've tried importing the spearfish_contours files via File-->Import Vector
> Map-->Multiple Formats Using OGR and get this message
>
> Invalid argument: can't _spawnl
> Unable to start driver <sqlite>
Could the problem be that \ and / slashes are used?
Hi all, [lunchtime; office PC hijacked for testing]
> > I've tried importing the spearfish_contours files via File-->Import
> > Vector Map-->Multiple Formats Using OGR and get this message
> >
> > Invalid argument: can't _spawnl
> > Unable to start driver <sqlite>
Markus Neteler wrote:
> Could the problem be that \ and / slashes are used?
Glynn:
Quite possibly. Try this:
...
+ G_convert_dirseps_to_host(dirpath);
I just tested this with the latest native installer and can reproduce
it in both the wx and tcl GUIs. (WinGRASS-6.4.0SVN-r39271-1-Setup.exe)
but.... I first ran v.buffer on roads@PERM and made a map in the local
mapset. I could query the roads_buff@user1 map ok, just not the roads@PERM
one in the non-current mapset, there I get a similar error but for dbf not
sqlite. For the local map I see the correct data output.
$GISDBASE is in C:\Documents and Settings\me\My Documents\GIS Database\
spearfish60\ or something like that with lots of spaces.
should this be using G_spawn() not _spawnl() now that we have a few
G_spawn() options to choose from?
Hamish
ps- the tcltk gui complains that it can't find the fontcap file, and
Font Selection from the config menu ends with a non-fatal tcltk error.
should this be using G_spawn() not _spawnl() now that we have a few
G_spawn() options to choose from?
No. G_spawn() uses the _P_WAIT flag, while the DBMI requires
_P_NOWAIT.
G_spawn_ex() supports background processes, but it doesn't support
redirecting descriptors on Windows. The only way to fix that is to
re-write it to use CreateProcess() rather than spawnvpe(); Python's
subprocess module would probably be a useful source of clues.
I'm a noob learning GRASS for a class. I've encountered a similar problem on
WinXP with the current install and I can't figure out what I'm doing wrong.
(Mon Oct 05 03:03:44 2009)
v.info -c streets_wake
Displaying column types/names for database connection of layer 1:
Invalid argument: can't _spawnl
ERROR: Unable to open driver <dbf>
(Mon Oct 05 03:03:44 2009) Command finished (0 sec)
What I know is that it occurs in both TCLTK and WxPython. It is specifically
related to the -c flag as other flags work fine. It isn't path or space
dependent - I tried with spaces and without on different drives and folder
locations. The dbf is not corrupted nor is it locked for some reason - I can
connect to it with ArcGIS and can open and edit it in excel. It isn't an
unregistered dll - I manually re-registered all the dlls in the GRASS
install folder. It isn't a user setting that I know of in the registry - I
manually deleted all orphan registry references to grass I could locate
after I uninstalled in an attempt to clear the problem earlier tonight.
I'm new to grass, not that familiar with the commands, and not sure where to
go next with this.
On Thu, Oct 1, 2009 at 7:31 PM, Glynn Clements <glynn@gclements.plus.com> wrote:
Markus Neteler wrote:
> I've tried importing the spearfish_contours files via File-->Import Vector
> Map-->Multiple Formats Using OGR and get this message
>
> Invalid argument: can't _spawnl
> Unable to start driver <sqlite>
Could the problem be that \ and / slashes are used?
Is this a harmless change? then I would submit it... otherwise it
will take years to figure this out unless someone actually compiles
and *tests* on Windows...
Submitting it would get it at least into the next test compilation.
I installed a previous version R38537-1 and attempted the exact same
connection as before under the exact same conditions. The connection was
successful and returned the correct values with the previous install where
as it returns errors under the current install.
Something has changed between 38537-1 and 39271-1 in a manner that breaks
the ability of the software to connect to dbfs.
>> Could the problem be that \ and / slashes are used?
>
> Quite possibly. Try this:
>
> --- lib/db/dbmi_base/dbmscap.c (revision 39364)
> +++ lib/db/dbmi_base/dbmscap.c (working copy)
> @@ -175,6 +175,7 @@
> dirpath = G_malloc(strlen("\"\\driver\\db\\\"")
> + strlen(G_gisbase()) + strlen(ent->d_name) + 1);
> sprintf(dirpath, "\"%s\\driver\\db\\%s\"", G_gisbase(), ent->d_name);
> + G_convert_dirseps_to_host(dirpath);
> #else
> G_asprintf(&dirpath, "\"%s/driver/db/%s\"", G_gisbase(), ent->d_name);
> #endif
Is this a harmless change? then I would submit it... otherwise it
will take years to figure this out unless someone actually compiles
and *tests* on Windows...
Submitting it would get it at least into the next test compilation.
I suggest committing it. As it stands, dirpath will contain a mix of
forward slashes and backslashes, which is the worst possible case
(there are two ways that it could break).
In the longer term, we should really try to get rid of forward slashes
on Windows altogether, rather than fixing the individual cases where
they don't work.
I report again a rounding problem, wondering why no one else seems to
have such problems.
Maybe someone con give me an good advice?
This time problem (see picture):
I have areas and lines and I want to remove the lines inside the areas-
I did:
-boundaries to lines ("arealines")
-patch lines and arealines
-break where they cross
-v.select -r -g operator=within
that should do, but due to ?roundings?, v.select does not do it
correct...(see picture) I would need a thresh in v.select, I guess,
hoping you might have an idea how to deal with it.
v.clean with snap does not work, because how can I tell the nodes at
line ends to snap to area boundaries.
On Tue, Oct 6, 2009 at 1:37 AM, Glynn Clements <glynn@gclements.plus.com> wrote:
Markus Neteler wrote:
...
Is this a harmless change? then I would submit it... otherwise it
will take years to figure this out unless someone actually compiles
and *tests* on Windows...
Submitting it would get it at least into the next test compilation.
