missing module named _textwrap

[hproe]

I have posted the support request below last week. I have had no reaction on it so far. Is there anything broken in the support request chain? Anyway, the error persists also in this week's builds.

-------- Original Message --------
Subject: [Org=EUMETSAT]: missing module
Date: Wed, 26 Jan 2011 03:10:15 -0600
From: HansPeter Roesli <satmet.hp@ticino.com>
To: mug@ssec.wisc.edu, satmet.hp@ticino.com

There appears to be a missing module in the most recent build. When e.g using
the Three Color (RGB) Image no image is displayed and the following error is
issued:
Traceback (innermost last):
File "<string>", line 16, in ?
ImportError: no module named _textwrap

[Jon]

Apologies for not being more forthcoming sometimes I forget the commit logs aren't public…maybe they should be? At any rate, I think I may have fixed this warning in every Jython context except formulas; mea culpa! I'll be sure to get a fix committed once I get to work.

FWIW I'm working on a series of changes to the "jython backend" within IDV/McV and so I'd like to pre-emptively apologize for any further issues!

That said, I'll be sure to test more thoroughly as well.

[joleenf]

Jon,

Since I ran into the same problem yesterday and also chose to submit the report through a form rather than on the forum, it would be nice to have some way to access both the commit logs and see the support form submissions. While some of this information may mean little to users, in these cases, it would help us.

Thanks,
Joleen

[Jon]

That's a good point! I can't think of a good solution while McV lingers in CVS…but if we had a publicly-readable SVN repository things would be pretty easy. I'll bring it up at our next team meeting!

FWIW my personal favorite solution is to build McV installers for each commit (immediately!) and storing the last N builds. It would significantly shorten turnaround time for you and make bug tracking a good deal easier.

[kbedka1]

I agree with Joleen, I've run into strange behaviors with nightly builds...I figured something was being worked on in that particular component of the code, but it would be nice to have a repository that verifies this.

[Jon]

Commits/Changelogs:

I should probably note that McV commits by themselves may not always be sufficient for determining the wonkiness of a given nightly build. The issue is that McV's underlying libraries (netcdf, visad, etc) are simply JAR files and are built independently of McV and without commit logs.

So you might see a McV commit message along the lines of "updated ncIdv.jar because of AMAZING_FEATURE_HERE" which causes McV to only display images/captions from unhappyhipsters.com. But even this situation is an improvement over the current set up.

_textwrap:

It took a little longer than anticipated, but I committed a tentative fix this afternoon. Please let me know if Jython misbehaves!

[joleenf]

Agreed, this is tricky. I believe the _textwrap message and an "initializing after unpersistence" error when trying to load an old bundle may have been related. I was trying to load a bundle with an isosurface colored by a parameter only to get the "initializing after unpersitence" error. In the _textwrap case, I was trying to create an isosurface colored by another parameter. I had believed these to be separate issues until I was able to solve both issues by reverting to an old build. Jay had mentioned that "things were funky" with the nightly around January 25, 2011. My point is that I would not likely have linked the two just looking at the commit logs or the support forms about "_textwrap." Even when Jay mentioned that the nightly was not working quite right, I wondered if it was the nightly or something had happened to the bundle. When the bundle loaded fine in an old build, it seemed to confirm Jay's idea.

[beckys]

Dave just added access to the version control commit messages for McIDAS-V, IDV, and VisAD to the McIDAS-V webpage - http://www.ssec.wisc.edu/mcidas/software/v/commits/

And I added a link to those commit messages from the McIDAS-V Download page - http://www.ssec.wisc.edu/mcidas/softwar ... nload.html

Like Jon said in an earlier post, the daily McIDAS-V updates are included in the nightly build of -V, but the IDV and VisAD updates are not automatically included in the nightly build of -V. They are included in -V on an "as needed" basis.

- Becky