[udig-devel] Interaction with system eclipse

[Tim Keitt]

Hi Adrian,

On 4/8/07, Adrian Custer <acuster at gmail.com> wrote:
> Hello Prof. Keitt,
>
> Glad you resolved your hexagonal tiling issue---it made me chuckle as
> being one of those really fundamental user needs, to re-partition a
> coverage, that none of our software yet handles with grace.
>
> The reports in your message have been filed as the following bugs:
>
> The first is not something that has come up recently but may be a known
> design issue that the refraction folk have chosen not to tackle yet---we
> will surely find out next week when they get back to work.
>     http://jira.codehaus.org/browse/UDIG-1231

I realize the 'extract to a single self-contained directory' approach
probably makes a lot of sense. It sure would be nice however if uDig
were packaged as a standard eclipse plugin. Then one could install via
the eclipse help menu. Better still would be some Ubuntu packages. I
think the complexity of installing uDig is inhibiting its wider
adoption. (Certainly in my case I've found it much easier to work with
QGIS.)

>
> The second is really unfortunate but surprises me because I've never run
> into similar problems. Would you be willing to report it with a bit more
> detail? I've got a bunch of different eclipses, of uDigs, and of
> workspaces on this machine so poking around to try to find the same
> issue as you experienced would be non-trivial. Is all of your eclipse
> and udig work occurring in the same workspace, say in ~/workspace?
> Eclipse and uDig have historically had the bad habit on gnu* systems of
> happily creating workspaces or writing into workspaces without asking
> the user for permission or for alternatives. So we may have to work
> around an eclipse issue or may simply need to make uDig more robust
> during startup. Thanks for any more info you could provide.
>     http://jira.codehaus.org/browse/UDIG-1232

I didn't think of the workspace directory. That's most likely the
culprit. I backed up my .eclipse directory prior to running uDig
(having been burned before), but did not think to move my workspace
out of the way. An easy test would be to copy the workspace directory,
run uDig and then run diff to compare the two workspaces.

>
>
> (Also, you probably saw Edzer Pebesma's response to my note on an R link
> to PostGIS. I'm hoping to get back to that discussion about how/if R
> wants to handle spatial analysis on objects significantly larger than
> main memory. However, I've run into a showstopper bug in uDig that I
> have to fix before I can get back to that interesting question.)

The approach I implemented in rgdal to handle large rasters in R could
easily be extended to vector (ogr) coverages (see my proposal on
r-sig-geo which is basically what Edzer is suggesting). I tried to
convince the 'sp' folks that we needed to be able to subclass the
spatial classes so that we could provide different data backends, but
I think that cannot be done as currently implemented. R could
certainly use a healthy dose of generic programming concepts.

>
>
> happy (my) birthday to all,

Cheers!

> --adrian
>
>
>

THK

-- 
Timothy H. Keitt, University of Texas at Austin
Contact info and schedule at http://www.keittlab.org/tkeitt/
Reprints at http://www.keittlab.org/tkeitt/papers/
ODF attachment? See http://www.openoffice.org/