The use of the name "GROUP" for file which stores the current imagery
group conflicts with the "group" directory on filesystems which
perform case folding.
Someone has added a fix specifically for Cygwin:
#ifdef __CYGWIN__
#define GROUPFILE "CURGROUP"
#else
#define GROUPFILE "GROUP"
#endif
#ifdef __CYGWIN__
#define SUBGROUPFILE "CURSUBGROUP"
#else
#define SUBGROUPFILE "SUBGROUP"
#endif
However:
1. The same problem may occur with MacOS X, where case-folding
filesystems are not uncommon.
2. Users of conventional Unices might access a GRASS database which is
stored on a case-folding filesystem, e.g. an SMB share which is used
by both Unix and Windows systems, or a FAT partition on a dual-boot
Linux/Windows box.
3. This creates a portability problem regarding distributed datasets
(e.g. the "imagery" dataset, which includes a GROUP file).
Consequently, I've changed the definitions so that the files are
always called CURGROUP and CURSUBGROUP.
Users will need to rename (or re-create) any GROUP/SUBGROUP files
after updating to the new code.
Also, any supplied datasets (e.g. "imagery") will need to be updated.
--
Glynn Clements <glynn.clements@virgin.net>
Glynn Clements wrote:
Consequently, I've changed the definitions so that the files are
always called CURGROUP and CURSUBGROUP.
Users will need to rename (or re-create) any GROUP/SUBGROUP files
after updating to the new code.
Also, any supplied datasets (e.g. "imagery") will need to be updated.
Glynn,
While I concur with your reasoning is it possible to do anything with the
code to improve backward compatibility? Perhaps if no CURGROUP directory
is found, the GROUP directory could be tried? I realize implementation may
be a little more complicated than that.
I humbly feel that changes like this can cause major user confusion and
inconvenience and that all reasonable efforts should be made to smooth the
compatability issues from the code side.
Finally, it would be nice to have a script that could be run to convert a
datastore to the new naming convention if backward compatibility can't be
maintained.
Best regards,
--
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up | Frank Warmerdam, warmerdam@pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush | Geospatial Programmer for Rent
Frank Warmerdam wrote:
> Consequently, I've changed the definitions so that the files are
> always called CURGROUP and CURSUBGROUP.
>
> Users will need to rename (or re-create) any GROUP/SUBGROUP files
> after updating to the new code.
>
> Also, any supplied datasets (e.g. "imagery") will need to be updated.
Glynn,
While I concur with your reasoning is it possible to do anything with the
code to improve backward compatibility? Perhaps if no CURGROUP directory
is found, the GROUP directory could be tried? I realize implementation may
be a little more complicated than that.
I humbly feel that changes like this can cause major user confusion and
inconvenience and that all reasonable efforts should be made to smooth the
compatability issues from the code side.
Finally, it would be nice to have a script that could be run to convert a
datastore to the new naming convention if backward compatibility can't be
maintained.
I've renamed the GROUP/SUBGROUP *files*, which record the *current*
group/subgroup, not the "group" directory (aka element), which stores
the actual group definitions.
AFAICT, the worst that will happen is that the user would have to
re-select (but not re-create) the current group/subgroup. That didn't
seem (to me) to be significant enough to warrant adding code to handle
a one-off situation.
I'm willing to add the code if it's deemed necessary. The only reason
why I didn't do that was a reluctance to add code which is almost
never used, not because of the (minimal) effort involved.
--
Glynn Clements <glynn.clements@virgin.net>
On Mon, Apr 15, 2002 at 09:34:28AM -0400, Frank Warmerdam wrote:
Glynn Clements wrote:
> Consequently, I've changed the definitions so that the files are
> always called CURGROUP and CURSUBGROUP.
>
> Users will need to rename (or re-create) any GROUP/SUBGROUP files
> after updating to the new code.
>
> Also, any supplied datasets (e.g. "imagery") will need to be updated.
Glynn,
While I concur with your reasoning is it possible to do anything with the
code to improve backward compatibility? Perhaps if no CURGROUP directory
is found, the GROUP directory could be tried? I realize implementation may
be a little more complicated than that.
I humbly feel that changes like this can cause major user confusion and
inconvenience and that all reasonable efforts should be made to smooth the
compatability issues from the code side.
Finally, it would be nice to have a script that could be run to convert a
datastore to the new naming convention if backward compatibility can't be
maintained.
Hi,
(finally back to Italy), I agree with Frank. At least we need a migration
script since some users may have many imagery groups (like us here).
Markus
A reminder:
Below GROUP/SUBGROUP changes still need to be described:
- in the NEWS.html file [I can do that]
- in the Programmer's manual [I can do that]
- for the imagery_dataset [I can do that]
Somewhere else?
Markus
On Sat, Apr 13, 2002 at 06:14:36AM +0100, Glynn Clements wrote:
The use of the name "GROUP" for file which stores the current imagery
group conflicts with the "group" directory on filesystems which
perform case folding.
Someone has added a fix specifically for Cygwin:
#ifdef __CYGWIN__
#define GROUPFILE "CURGROUP"
#else
#define GROUPFILE "GROUP"
#endif
#ifdef __CYGWIN__
#define SUBGROUPFILE "CURSUBGROUP"
#else
#define SUBGROUPFILE "SUBGROUP"
#endif
However:
1. The same problem may occur with MacOS X, where case-folding
filesystems are not uncommon.
2. Users of conventional Unices might access a GRASS database which is
stored on a case-folding filesystem, e.g. an SMB share which is used
by both Unix and Windows systems, or a FAT partition on a dual-boot
Linux/Windows box.
3. This creates a portability problem regarding distributed datasets
(e.g. the "imagery" dataset, which includes a GROUP file).
Consequently, I've changed the definitions so that the files are
always called CURGROUP and CURSUBGROUP.
Users will need to rename (or re-create) any GROUP/SUBGROUP files
after updating to the new code.
Also, any supplied datasets (e.g. "imagery") will need to be updated.
--
Glynn Clements <glynn.clements@virgin.net>
_______________________________________________
grass5 mailing list
grass5@grass.itc.it
http://grass.itc.it/mailman/listinfo/grass5