[Gdal-dev] Re: GDAL/OGR updates: configure and .cvsignore

[Frank Warmerdam]

Markus Neteler wrote:
> Frank,
> 
> find attached a few patches:
> 
> cvsignore_update.tar.gz
> - updates for *.lo 
>   (
>    ogr/ogrsf_frmts/o/.cvsignore
>    frmts/o/.cvsignore
>    apps/.cvsignore
>    frmts/hfa/.cvsignore
>    )
>   note: two files are new.

Markus,

I have committed the .cvsignore updates.  I see you were primarily
adding .lo files to the list.  I would suggest you add these to your
.cvsrc file as a more comprehensive approach.  For instance, my .cvsrc
reads:

update -I *.class -I *.lo -I .libs -I *.la

This ensures it ignores Java .class files, libtool .lo files, any
libtool .libs directories and any libtool .la files.

> - maybe more relevant:
> I have extended configure.in/aclocal.m4 to print out the configuration
> at end of the 'configure' run:
> [...]
> configure: creating ./config.status
> config.status: creating GDALmake.opt
> config.status: creating port/cpl_config.h
> config.status: port/cpl_config.h is unchanged
> 
> GDAL is now configured for i686-pc-linux-gnu
> 
>   Installation directory:    /usr/local/grass
>   C compiler:                gcc -O2
>   C++ compiler:              g++ -O2
> 
>   LIBTOOL support: yes
>   LIBZ support: yes
> 
>   GDAL formats:   aigrid bmp bsb ceos ceos2 dted elas envisat gdb gtiff gxf hdf4 hfa iso8211 jdem l1b mem nitf ogdi pcidsk pg raw sdts shapelib usgsdem vrt xpm
> 
>   OGR support:    yes
>   OGR formats:    avc dgn generic gml mem mitab ntf ogdi pg rec s57 sdts shape tiger
> 
> We have that in GRASS, it's also in R-stats and might be convenient
> also for GDAL/OGR.
> 
> Find attached the patches against CVS:
> diffs_aclocal.m4_configure.in.diff
> (configure.in and aclocal.m4)

This is a fantastic idea; however, I have a few concerns about the execution.
One problem is that the GDAL_FORMATS macro in GDALmake.opt isn't really a list
of drivers, but rather a list of directories to recurse into.  So, iso8211
for instance isn't a GDAL format at all.  On the other hand, all the "raw"
formats are missing.  These problems are shared by the "gdal-config --formats"
report which is also based on the GDALmake.opt GDAL_FORMATS list.

My other concern is that you have had to add alot of stuff in configure.in about
the various format options to generate the report.  My concern is that this is
going to be a bother to maintain, which means it likely won't be well maintained
and so it will be inaccurate.  (At least I know my own limitations!)

So, I like the idea of a summary report, and I am willing to commit the part
about the compiler, libtool support and so forth.  I am less keen to try and
report a full set of drivers at this point - for GDAL or OGR.  If we do choose
to report them, perhaps it should just be based on invoking the makefile with
a special option to report the drivers configured as the makefile knows them?
Then we don't have to maintain a big chunk of extra logic in the configure.in.

I am opening this up to broader input.  Thanks for the patches!

Best regards,

-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent