[GRASS-dev] [GRASS GIS] #70: i.target from GUI: strip @mapset part

GDAL · February 27, 2008, 2:08am

#70: i.target from GUI: strip @mapset part
---------------------+------------------------------------------------------
Reporter: hamish | Owner: grass-dev@lists.osgeo.org
Type: defect | Status: new
Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Keywords: |
---------------------+------------------------------------------------------
Hi,

if you call i.target from a GUI and select a group it appends @mapset. If
you run it like that it takes the full name literally and instead of
updating the group's target it creates a new group of name "group@mapset".
Either i.target or I_put_target() should check that if the @mapset part is
given it refers to the current mapset, then strip off the @mapset part.

Otherwise you end up with
{{{
GRASS> g.list group

group
group@mapset
group@mapset@mapset

etc.
}}}

I changed the group option gisprompt from old,group,group to any,gr,gr.
This makes the GUI group picker button go away, but perhaps the GUI picker
button should be coded to remain in that case?

Or to solve this should we try changing the gisprompt to mapset,gr,gr so
it calls G_ask_in_mapset()?

Hamish

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70>
GRASS GIS <http://grass.osgeo.org>
GRASS Geographic Information System (GRASS GIS) - http://grass.osgeo.org/

GDAL · February 27, 2008, 12:30pm

#70: i.target from GUI: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords:
----------------------+-----------------------------------------------------
Comment (by glynn):

Replying to [ticket:70 hamish]:

> if you call i.target from a GUI and select a group it appends @mapset.
If you run it like that it takes the full name literally and instead of
updating the group's target it creates a new group of name "group@mapset".
Either i.target or I_put_target() should check that if the @mapset part is
given it refers to the current mapset, then strip off the @mapset part.

> I changed the group option gisprompt from old,group,group to any,gr,gr.
This makes the GUI group picker button go away, but perhaps the GUI picker
button should be coded to remain in that case?
>
> Or to solve this should we try changing the gisprompt to mapset,gr,gr so
it calls G_ask_in_mapset()?

To solve this, i.group needs to be modified to handle the @mapset part
correctly. That is the only course of action which can accurately be said
to *solve* the issue.

There may be workarounds, but a workaround isn't the same thing as a
solution.

And any function named G_ask_<something> shouldn't be used for anything.
Those functions cause problems for GUI use (requiring a terminal), and
will disappear the moment that 7.x comes into being.

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:1>
GRASS GIS <http://grass.osgeo.org>
GRASS Geographic Information System (GRASS GIS) - http://grass.osgeo.org/

GDAL · March 6, 2008, 6:04am

#70: i.target from GUI: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords:
----------------------+-----------------------------------------------------
Comment (by hamish):

i.class too apparently:
http://thread.gmane.org/gmane.comp.gis.grass.user/22882

Hamish

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:2>
GRASS GIS <http://grass.osgeo.org>
GRASS Geographic Information System (GRASS GIS) - http://grass.osgeo.org/

GDAL · March 30, 2008, 4:13am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Changes (by hamish):

  * keywords: => imagery
* cc: Paula.Sankelo@fimr.fi (added)
  * summary: i.target from GUI: strip @mapset part => imagery modules:
              strip @mapset part

Comment:

i.group, i.target modified in SVN to handle @mapset qualified names in
r30802.
i.class done in r30803.

As this is a systemic problem a better approach would be to improve the
imagery libs to deal with that -- renaming bug and leaving it open.

I don't really use the imagery modules so I'm not sure what other modules
are badly crippled from this. Proably most, but as mentioned above the way
to tackle that is to fix it properly in the imagery lib fns.

e.g. listing maps contained in a group from another mapset should work.

Hamish

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:3>
GRASS GIS <http://grass.osgeo.org>
GRASS Geographic Information System (GRASS GIS) - http://grass.osgeo.org/

GDAL · March 30, 2008, 9:37am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by glynn):

Replying to [comment:3 hamish]:
> As this is a systemic problem a better approach would be to improve the
imagery libs to deal with that -- renaming bug and leaving it open.

I have attached a [http://trac.osgeo.org/grass/attachment/ticket/70/
imagery-find.patch patch] which should fix the I_find_* functions to
handle qualified names. Can someone test it?

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:4>
GRASS GIS <http://grass.osgeo.org>
GRASS Geographic Information System (GRASS GIS) - http://grass.osgeo.org/

GDAL · April 8, 2008, 10:39pm

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by glynn):

Replying to [comment:4 glynn]:

I have attached a more extensive version of the previous patch.

attachment:ticket:70:imagery-mapset.patch

This should cover all of the cases where filenames were being constructed
manually.

There remains the issue that the imagery library can only handle groups in
the current mapset, for both read and write. Most of that could be fixed
by replacing G_mapset() with "". However, the I_find_* functions really
need _new and _old versions, as (unlike the G_find_* functions) they don't
accept a mapset argument.

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:5>
GRASS GIS <http://grass.osgeo.org>
GRASS Geographic Information System (GRASS GIS) - http://grass.osgeo.org/

Glynn_Clements1 · April 12, 2008, 8:53pm

GRASS GIS wrote:

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by glynn):

Replying to [comment:4 glynn]:

I have attached a more extensive version of the previous patch.

attachment:ticket:70:imagery-mapset.patch

This should cover all of the cases where filenames were being constructed
manually.

As there were no objections, I have committed this.

It shouldn't be necessary for imagery modules to manually strip the
@mapset part (unless they are processing map names themselves).

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

GDAL · May 18, 2008, 9:18pm

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by neteler):

It appears that i.group is now broken:

http://lists.osgeo.org/pipermail/grass-user/2008-May/044688.html

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:6>
GRASS GIS <http://grass.osgeo.org>

GDAL · May 19, 2008, 9:15am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by neteler):

(i.group functional again).

I tried to use i.class, NC data set:

{{{
i.group mylsat7_2000 sub=mylsat7_2000
in=lsat7_2000_10,lsat7_2000_20,lsat7_2000_30,lsat7_2000_40,lsat7_2000_50,lsat7_2000_61,lsat7_2000_70,lsat7_2000_80

g.region rast=lsat7_2000_40 -p
d.mon x0
i.class map=lsat7_2000_40 group=mylsat7_2000 subgroup=mylsat7_2000
outsig=lsat7_2000_sig

RESULT SIGNATURE
WARNING: unable to create signature file lsat7_2000_sig for subgroup
mylsat7_2000 of group mylsat7_2000
ERROR: Unable to open output signature file 'lsat7_2000_sig'
}}}

It seems to be an issue in lib/imagery/sigfile.c (devel_grass6 branch).

Markus

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:7>
GRASS GIS <http://grass.osgeo.org>

GDAL · May 19, 2008, 9:37am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by glynn):

Replying to [comment:7 neteler]:

> (i.group functional again).
>

{{{
> I tried to use i.class, NC data set:
> WARNING: unable to create signature file lsat7_2000_sig for subgroup
> mylsat7_2000 of group mylsat7_2000
> ERROR: Unable to open output signature file 'lsat7_2000_sig'
}}}

> It seems to be an issue in lib/imagery/sigfile.c (devel_grass6 branch).

Hopefully fixed in r31419 (trunk), r31420 (6.4).

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:8>
GRASS GIS <http://grass.osgeo.org>

GDAL · May 19, 2008, 10:12am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: minor | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by neteler):

Possibly we need yet another one? Patch attached.

Markus

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:9>
GRASS GIS <http://grass.osgeo.org>

GDAL · May 19, 2008, 10:14am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: major | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Changes (by neteler):

* priority: minor => major

Comment:

i.class is not yet working.

Markus

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:10>
GRASS GIS <http://grass.osgeo.org>

GDAL · May 20, 2008, 7:20am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: major | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by neteler):

Fix applied for i.class (6.4 branch: r31439 and trunk: r31440). Working
again.

Markus

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:11>
GRASS GIS <http://grass.osgeo.org>

GDAL · May 21, 2008, 9:07am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: major | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by neteler):

While i.class now works from cmd line, it still fails interactively:

{{{
GRASS 6.4.svn (nc_spm_07):~ > i.class

OPTION: Name of raster map to be displayed
key: map
format: name
required: YES

Enter the name of an existing raster file
Enter 'list' for a list of existing raster files
Hit RETURN to cancel request
> lsat7_2002_40
<lsat7_2002_40>

OPTION: Name of input imagery group
key: group
format: name
required: YES

Enter the name of an existing group file
Enter 'list' for a list of existing group files
Hit RETURN to cancel request
> mylsat7_2000
<mylsat7_2000>

OPTION: Name of input imagery subgroup
key: subgroup
format: name
required: YES

Enter the name of an existing subgroup file
Enter 'list' for a list of existing subgroup files
Hit RETURN to cancel request
> list
<list>
----------------------------------------------

no subgroup files available in current mapset
----------------------------------------------

Enter the name of an existing subgroup file
Enter 'list' for a list of existing subgroup files
Hit RETURN to cancel request
> mylsat7_2000
<mylsat7_2000>

** mylsat7_2000 - not found **

Enter the name of an existing subgroup file
Enter 'list' for a list of existing subgroup files
Hit RETURN to cancel request
>
}}}

I suspect a bug in [source:grass/trunk/lib/imagery/find.c]

Markus

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:12>
GRASS GIS <http://grass.osgeo.org>

GDAL · May 21, 2008, 11:41am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: major | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by glynn):

Replying to [comment:12 neteler]:

> While i.class now works from cmd line, it still fails interactively:

That's a separate issue, which was introduced in r24069.

> <list>

> no subgroup files available in current mapset

If you read that literally, it should give you a clue what's going on
here.

> I suspect a bug in [source:grass/trunk/lib/imagery/find.c]

Nope; it's a limitation of G_parser(), which cannot handle subgroups.
Essentially, "old,subgroup,subgroup" isn't a valid ->gisprompt setting.

G_parser() itself knows nothing of the imagery library, and there is no
way for individual modules to extend it. When G_parser() performs
interactive input for an option with "old,<element>,<description>" as the
->gisprompt value, it just checks for the file in the <element> directory.
Similarly, if you type "list", it just lists the <element> directory.

Well, mapsets don't have a "subgroup" directory, as subgroups belong to
specific groups, not the mapset. Beyond that, G_parser() would have no way
of knowing which group the subgroup belongs to (it cannot know that the
value of the group= option indicates this).

Ultimately, you'll have to just remove the ->gisprompt setting, leaving
the option as just a string.

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:13>
GRASS GIS <http://grass.osgeo.org>

GDAL · May 21, 2008, 1:22pm

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: major | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by neteler):

Indeed, removing for subgroup makes it pass. Fixes in:

GRASS 6.3.svn: r31465

GRASS 6.4.svn: r31464

GRASS 7.0.svn (trunk): r31463

Now i.class works again interactively, too.

Markus

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:14>
GRASS GIS <http://grass.osgeo.org>

GDAL · June 3, 2008, 11:14am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: major | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by neteler):

It seems that i.atcorr has some mapset related problems, too:

{{{
GRASS 6.4.svn (nc_spm_07):~ > g.gisenv
GISDBASE=/home/neteler/grassdata
LOCATION_NAME=nc_spm_07
MAPSET=sqlite
DEBUG=0
GRASS_GUI=text

GRASS 6.4.svn (nc_spm_07):~ > g.mapsets -p
sqlite PERMANENT

GRASS 6.4.svn (nc_spm_07):~ > i.atcorr -b iimg=lsat5_1987_60@landsat
iscl=-6.978740,191.600 icnd=ddddd oimg=test_p112r056_b1.rad.atcorr
oscl=0,255
G__open(r): mapset (sqlite) doesn't match xmapset (landsat)
G__open(r): mapset (sqlite) doesn't match xmapset (landsat)
WARNING: Unable to open header file for raster map
<lsat5_1987_60@landsat@sqlite>
ERROR: Unable to retreive header dat for input image

GRASS 6.4.svn (nc_spm_07):~ > g.mapsets add=landsat

GRASS 6.4.svn (nc_spm_07):~ > i.atcorr -b iimg=lsat5_1987_60
iscl=-6.978740,191.600 icnd=ddddd oimg=test_p112r056_b1.rad.atcorr
oscl=0,255
WARNING: Unable to open header file for raster map <lsat5_1987_60@sqlite>
ERROR: Unable to retreive header dat for input image
}}}

Markus

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:15>
GRASS GIS <http://grass.osgeo.org>

GDAL · June 3, 2008, 11:36am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: major | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by neteler):

OK, the i.atcorr mapset problems were unrelated. Fixed in r31614 and
r31615.

Markus

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:16>
GRASS GIS <http://grass.osgeo.org>

GDAL · June 21, 2008, 8:32am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: major | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by neteler):

I have found a new related bug: -r of i.group fails.

{{{
g.gisenv set=DEBUG=3
GRASS 6.4.svn (nc_spm_07): > i.group -r mylsat7_2000 sub=mylsat7_2000
in=lsat7_2000_61
D3/3: remove_subgroup_files: ref_tmp.nfiles 8
D3/3: tmp_name lsat7_2000_61, ref_tmp.file[0].name: lsat7_2000_10
D3/3: mapset neteler, ref_tmp.file[0].mapset: landsat
D3/3: tmp_name lsat7_2000_61, ref_tmp.file[1].name: lsat7_2000_20
D3/3: mapset neteler, ref_tmp.file[1].mapset: landsat
D3/3: tmp_name lsat7_2000_61, ref_tmp.file[2].name: lsat7_2000_30
D3/3: mapset neteler, ref_tmp.file[2].mapset: landsat
D3/3: tmp_name lsat7_2000_61, ref_tmp.file[3].name: lsat7_2000_40
D3/3: mapset neteler, ref_tmp.file[3].mapset: landsat
D3/3: tmp_name lsat7_2000_61, ref_tmp.file[4].name: lsat7_2000_50
D3/3: mapset neteler, ref_tmp.file[4].mapset: landsat
D3/3: tmp_name lsat7_2000_61, ref_tmp.file[5].name: lsat7_2000_61
D3/3: mapset neteler, ref_tmp.file[5].mapset: landsat
D3/3: tmp_name lsat7_2000_61, ref_tmp.file[6].name: lsat7_2000_70
D3/3: mapset neteler, ref_tmp.file[6].mapset: landsat
D3/3: tmp_name lsat7_2000_61, ref_tmp.file[7].name: lsat7_2000_80
D3/3: mapset neteler, ref_tmp.file[7].mapset: landsat
D1/3: writing subgroup REF
D2/3: G__home home = /home/neteler
WARNING: No raster map removed
i.group complete.
}}}

It compares the wrong mapsets here (G_mapset() should possibly not
used on the left side, see function remove_subgroup_files() of main.c of
i.group.

Markus

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:17>
GRASS GIS <http://grass.osgeo.org>

GDAL · June 21, 2008, 9:44am

#70: imagery modules: strip @mapset part
----------------------+-----------------------------------------------------
  Reporter: hamish | Owner: grass-dev@lists.osgeo.org
      Type: defect | Status: new
  Priority: major | Milestone: 6.4.0
Component: default | Version: svn-trunk
Resolution: | Keywords: imagery
----------------------+-----------------------------------------------------
Comment (by glynn):

Replying to [comment:17 neteler]:
> I have found a new related bug: -r of i.group fails.

Do r31775 and r31776 help?

G_mapset() returns a pointer to a static buffer. If you modify this
buffer, you modify the return value for all subsequent G_mapset() calls.
It should probably be change to "const char *" to catch this.

--
Ticket URL: <http://trac.osgeo.org/grass/ticket/70#comment:18>
GRASS GIS <http://grass.osgeo.org>