Today's Work

In the morning I wrote a script that checks LDAP for missing fields and sends an email alert about what it's found. Then I tried cleaning up the CVS mess again for viewExpired and told Mike to move it to test again. I added some code to the phonebook LaTex file to prevent bad page breaks. And in the afternoon I work on an email link obfuscator for Mike Wiggs.

Today's Work

I finished looking at the copy bug, wrote up a mail message and sent it to the web development list. This is call 375387 The DVAs reported an isuue with my updates script. I investigated, but couldn't find any problem. I also got a request to increase the line spacing in the phonebook, which I did. Mike reported problems doing an update from CVS on test, but I couldn't find any problems.

Today's Work

Doing a cut operation in the top level directory leads to the following result:

perl webhead.pl http://www.stsci.edu:8082/?manage_copyObjects:method=Copy
==> http://www.stsci.edu:8082/?manage_copyObjects:method=Copy
==> http://www.stsci.edu:8082/resources/?manage_copyObjects:method=Copy
HTTP/1.1 200 OK
Date: Mon, 29 Dec 2003 19:48:17 GMT
Server: Zope/(Zope 2.6.0 (source release, python 2.1, linux2), python 2.1.3, linux2) ZServer/1.1b1
Cache-Control: proxy-revalidate
Content-Type: text/html
Etag: 
Content-Length: 627
X-Cache: MISS from www.stsci.edu
Connection: close

Here's the problem. The object that invokes the copy method is the folder currently listed in manage_main. The object must be a folder because only folders support copying (CopyContainer is a base class only for ObjectManager). But the root folder for www.stsci.edu gets redirected to the resources page.

I've switched over to the new version of themer, make depend is still borked, though.

Today's Work

I fixed up CVS and resubmitted the change to view expired news items. This change is call 384677.

I also worked on the copy bug, which is call 375387. I recreated the problem on 8082 by pressing the copy button in the root folder of www.stsci.edu/manage. I looked at the Zope logs and saw that Zope is receiving a request for the www.stsci.edu/resources. No big surprise so far.

Looking at the source for the management interface showed that the copy button calls manage_copyObjects, which is in OFS.CopySupport.py. It's a simple routine that sets a cookie and then calls manage_main. I hunted for the source of manage_main and found it in OFS/dtml/main.dtml. It in turn pulls in pieces of dtml from App/dtml There's nothing I can see in any of the code that would cause the copy problem. I think there's a redirect problem, although I can't see what it might be. I'll have to look at the CWS logs to see what's going on.

Today's Work

I resolved a problem with the leave system that turned out to be a data entry mistake in LDAP. This was call 385228. I worked on a Zope corruption problem from Melanie Strawbridge and sent a suggestion to Mike Watterson. This was call 385412.

Today's Work

I answered a question about ftp into Zope (CNSHD384493). I helped Jim Younger with a problem with the counter on the orient script that mysteriously went away. I wrote up a test plan for the view expired fix for STObjects and committed the changes to cvs (CNSHD384677). I retested Harry's LDAP manager changes (CNSHD384270).

Today's Work

I tested Harry's new version of the ldap manager and it had javascript problems. I did the ldap presentation for tomorrow.

Today's Work

Mike reported the orient script showed broken images. It turned out that the slash separating the image name from the rest of the url was missing. I have no idea why it went missing. I looked at solving the problem of writing a script to give everyone with auditor role management privileges. Jim suggested that it wasn't necessary and that we can use the admin account to access the mirrored database.

Today's Work

I worked with Jim to help resolve the problem with the production instance of Zope sending out so many error messages. We looked at the code and determined the problem was originating with the ZEO server and probably was a problem with a corrupted databse. Mike ran the 2.6.2 version of fsrestore on the database, which seemed to resolve the problems. After that I tried to make the orient script taint safe. I wound up writing a globbing function called safe_glob that didn't cause taint checking to throw an error.

Today's Work

I spent the day working on the orient script, trying various things to get it to work. I also took the Unix test for root privileges and passed it.