Format translator development

[Michel Wurtz - ENGEES/CEREG]

Rich Shepard wrote:
> David,
> 
>   You may be correct. About a decade ago, we were discussing this on FIDOnet.
> What we *don't* want to use is SDTS; that's a complex pig whose only
> function is to provide careers for some government bureaucrats.
>
>   I haven't any opinion regarding some neutral, generic translator. But, on
> a practical note, how many translators do we need? Two (perhaps three) from
> ESRI, one each from MapInfo, AutoCAD, and SDTS. IIRC, other GISes (Pamap,
> ERDAS, TNTMips, Idrisi, EPPL-7) can write to one of the above formats.

Richard,

I agree with you.  There is some similar French "neutral format".
Since it must deal with anything possible in GIS formats, it's not
very usable. I think that we should write only a few "good" translator
for well known GISes.  Any "serious" product can read/write them.

>   The ARC/Info .e00 translators apparently work (in 5.x), but are
> inconsistent in 4.x; ARCView Shape, SDTS and (now) MapInfo translators are
> being developed. We're almost there. On the other hand, since the GRASS data
> format is changing between 4.x and 5.x, there's merit in some intermediate
> format. With that, future GRASS format changes can be more easily
> accommodated.

Even when Grass formats change, the API stay the same, or at least is not
too different.  Updating m.in.e00 from 4.3 to 5.0 was not a too difficult task
for what was written.  Including the new functionnalities (floating point
raster, ...) and debugging them was the most time-consuming task.

Well, I just looked at the last version of grass4.3 and it seems that the last
bug corrections I sent were lost.  I don't add new functionnalities to 4.3,
but  try to correct the bugs in both versions.
I will resent the complete package to Bruce so I hope the next version of 4.3
will be accurate.  In the meantime, I can send directly the complete package
for 4.3.  Some good news is that e00b2a is no longer needed as both versions
use the same e00 read function, that can manage compressed e00 files.
As the author (and maintener) of m.in.e00, I am always interested (if not happy
:-)
if something goes wrong when importing e00 files, so don't hesitate to tell
me when this occurs.

BTW, if my experience with m.in.e00 can help you with a mif/mid importer,
I will be glad to contribute (I just asked myself "what about Mapinfo
maps in Grass ?" yesterday).
Shapefile are probably closer from MIF/MID that e00 files, but on the other
hand, the Grass internal vector format is quite close from arc/info native
format (arc shared by adjacent polygons, point inside a polygon for identifying
it).  I'm sure that some functions will be needed by the import programs and may
be usefull elsewhere : for example computing a polygon "centroid" to convert
shapefile or MFI/MID description of polygons to the Grass one.

Best regards
-- 
Michel Wurtz    ENGEES - CEREG
                1, quai Koch - BP 1039, F-67070 STRASBOURG cedex
                Tel: +33 03.88.24.82.45  Fax: +33 03.88.37.04.97