WinGRASS GIS.m: cannot select maps from mapset GUI

[hamish]

Hi,

GIS.m on wingrass refuses to let you select maps from another mapset.

e.g. spearfish, from the user1 mapset:

• start gis.m
• click the button to add a new raster map
• click the button at the left of the map name entry box to open the map list GUI
• only the user1 folder is open, the PERMANENT folder will not open.
• type elevation.dem@PERMANENT by hand and it works.

?, Hamish

[hamish]

ah, the little [+] [-] boxes to expand the folders are missing on wingrass for some reason. so the PERMANENT folder stays permanently closed.

[hamish]

you can't select maps from the current mapset either. see #637

[marisn]

I hope I have fixed this issue in 6.5. r38295

Please test it on Linuxes, Macs and Windblows. If OK, needs to be merged into 6.4.0.

[hamish]

Thanks a lot Maris! Fingers crossed..

I am not well set up for building WinGRASS or even dropping in new tcl files. If someone who is can test 6.5svn and report back please.... ?

Picking a map in gis.m still seems to work ok work on Linux.

cheers, Hamish

[hellik]

Replying to hamish:

Thanks a lot Maris! Fingers crossed..

I am not well set up for building WinGRASS or even dropping in new tcl files. If someone who is can test 6.5svn and report back please.... ?

Picking a map in gis.m still seems to work ok work on Linux.

cheers, Hamish

hi

grass65-svn compiled (in osgeow4-tree) at rev38296

now it's possible to select raster and vector in the tcltk-gui. so the ticket should be closed.

all the best Helli

[hamish]

In 6.5svn r38310 I set MAPSET back to a normal string -- it's not a multipart path name. It seemed odd/wrong to add "pwd/" to it then strip it away again a few lines later.

also in select.tcl line 180,

source:grass/branches/develbranch_6/lib/gtcltk/select.tcl@38310#L180​

should $path be "quoted"?

Hamish

[hamish]

Replying to hellik:

grass65-svn compiled (in osgeow4-tree) at rev38296 now it's possible to select raster and vector in the tcltk-gui.

can you select a different vector points symbol (say something from the geology folder) and display it correctly?

Hamish

[hamish]

(leave open until everything is backported & job is done)

[marisn]

Thanks Hamish for spotting my oversight with mapset.

$path in select.tcl seems to work just fine without quotes (no idea why - must be something of tcl beloved lists vs strings), still there was another issue with running anything if GISBASE contains spaces. Fixed in r38311. Also an candidate for 6.4.0.

Leaving open till it's tested and backported.

[cmbarton]

Works fine in GRASS 6.4 and 6.5 svn on Mac OS X.

Michael

[hamish]

Replying to cmbarton:

Works fine in GRASS 6.4 and 6.5 svn on Mac OS X.

yeah, it's only on WinGrass. I think Maris committed a fix for this in 6.5svn, but it awaits review & backporting by someone who understands Tcl better than I do.

Hamish

[cmbarton]

I can't test this on windows unless I have a binary to give to one of my students who uses windows. I've been out of the loop for awhile. But if Maris can give me the fix, I can manually insert it into a 6.4 Windows binary and see if it works.

Michael

[hellik]

grass65 selfcompiled in the osgeo4w-stack on WinVista32 rev39115 nc-dataset mapset: user1

there's no problem to choose a vector or raster from another mapset like PERMANENT in the nc-dataset. i've tried this with spaces (C:\gis data\grassdata) and without spaces (C:\gisdata\grassdata) in the path to the location.

[hellik]

Replying to hellik:

grass65 selfcompiled in the osgeo4w-stack on WinVista32 rev39115 nc-dataset mapset: user1

there's no problem to choose a vector or raster from another mapset like PERMANENT in the nc-dataset. i've tried this with spaces (C:\gis data\grassdata) and without spaces (C:\gisdata\grassdata) in the path to the location.

tested in the tcltk/gis.m GUI

[cmbarton]

Is this relevant anymore? If the problem is

1) with the TclTk GUI and not wxpython, 2) only in Windows, and 3) if wxpython is the default GUI for WinGRASS 6.4,

do we need to worry about it?

Michael

[hamish]

Replying to cmbarton:

Is this relevant anymore?

To me, yes. Others are of course free to decide that for themselves and spend their time worrying about issues they find more interesting, as they like.

bugs we know about should be fixed. especially ones that cripple an entire subsystem. anyway, this is just a small backport, not a huge investment in/redirection of effort. The larger problem on spaces in path names is of course another matter. I'm still a bit frustrated that the MSys devs won't even take patches for this class of bug.

I would also note that making things more robust on WinGrass typically has the byproduct of making thinks slightly more robust on other platforms as well.

Hamish

[cmbarton]

If it's just a backport, I'm definitely in favor doing this.

Like I said, I can probably stick a TclTk patch into a student's 6.4 Windows binary to test this if it is a help. To do this on Windows, however, it will be much easier if I get the entire patched tcltk module instead of a patch file.

Michael

[marisn]

Please test r39149.

[hamish]

tested in latest WinGrass 6.4 installer (WinGRASS-6.4.0SVN-r39271-1-Setup.exe).

works; closing bug.

(it complains that it can't find the fontcap file though)

thanks, Hamish