Monday, October 15, 2007

Repository key expired III

The repository archive key had expired at October 10th 2007. The new expire date is October 15th 2008. You will have to update the key in your apt keyring. The updated key can be found at pgp.mit.edu keyserver or locally in ASCII format at wgdd_archive_key.asc. Detailed information can be found in section Archive signing key.

Monday, June 18, 2007

Gabedit, DRAWxtl and libint are on their way into Debian

This is an information for all those users and developers, which are interested in chemistry and computational chemistry packages in Debian: As part of the debichem project at Alioth, I recently added Gabedit, DRAWxtl and libint to our project SVN repository. All packages are in a fairly good state for an upload into the Debian NEW queue. Today we (Michael Banck and me) began with gabedit. The other packages will follow soon (I swear).

Also thanks to Abdul-Rahman Allouche, Larry Finger (and the other DRAWxtl authors) and Edward Valeev for these software packages and their response to my mails :)

Our SVN repository further contains Debian packaging files for the MOLDEN software package. Unfortunately we are currently not allowed to ship the MOLDEN source or binaries within Debian. However, you can use our SVN repository to build the package for yourself. Just install the svn-buildpackage package, download the MOLDEN source tarball and our SVN (wnpp/molden/), configure svn-buildpackage for this setup (a complete how-to will follow soon) and then build the package by running svn-buildpackage.

If you are interested in our work and/or if you want to help, subscribe to our mailing list and meet us there.

Monday, May 21, 2007

gURLChecker accepted into Debian

gURLChecker has been officially packaged for the Debian distribution. I'm therefor going to remove my own package of this piece of software from my repository. Updating to the official Debian packages should work without problems.

Monday, May 14, 2007

Etch, sftp and the rssh shell

And finally I discovered an issue. SFTP did not work anymore. The debug session showed:

[..]
debug1: Sending subsystem: sftp
debug1: client_input_channel_req: channel 0 rtype exit-status reply 0
debug1: channel 0: free: client-session, nchannels 1
debug1: Transferred: stdin 0, stdout 0, stderr 0 bytes in 0.2 seconds
debug1: Bytes per second: stdin 0.0, stdout 0.0, stderr 0.0
debug1: Exit status 0

The user only has the RSSH shell (sftp enabled) and he is further limited by the following entry in .ssh/authorized_keys:

command="/usr/lib/sftp-server" [SSH-key]

But this doesn't work anymore. /usr/lib/sftp-server now is a symlink to /usr/lib/openssh/sftp-server and I had to change the .ssh/authorized_keys for the user to:

command="/usr/lib/openssh/sftp-server" [SSH-key]

and access is granted again.

Update

This one can be found in syslog:

rssh[xxx]: user XXX attempted to execute forbidden commands
rssh[xxx]: command: /usr/lib/sftp-server

Sarge -> Etch transition

Yesterday I've finally updated my server from Sarge to Etch (around 300 packages to update, 100 newly installed, 15 to remove). The whole transition went smooth in a screen session in about 4 hours or so.

  1. I began with a normal upgrade as suggested by the release notes.
  2. The next step introduced the new initrd-tools and libc6 to the system ...
  3. ... followed by the installation of a new linux-image and udev.
  4. The last step in updating the packages involved a complete dist-upgrade.
  5. Then it was time for a reconfiguration of all the services, that installed massively changed config files [1] and fixing all the chroots and broken package configurations caused by package transitions.
  6. The reboot brought up the system at first go. Checking with dmesg and bootlogd did not show any serious issues: /dev ok and all services up and running with the new kernel. So besides a few modules, which are not longer available or changed their name, no issues occured.
  7. With the new system up and running I did some cleanup, removing unused packages and those, that were removed in Etch (all together around 25).

That's all. New system up and running (with a few configuration works left) and no need to restore the system from the image :). So yes, I'm satisfied. Let's see, if issues will occur later.

[1] In this case I installed the package maintainers version and re-added my changes later.

Thursday, February 15, 2007

Let file(1) recognize a chemical MIME type - the next level

As written earlier, the file command may detect chemical MIME types, if you feed its database with the necessary definition rules. Now check out the cmd.magic.mime file and run the file command with

$ file -m cmd.magic.mime -i your_test_file.ext

for one or more chemical files. The detection is limited to chemical MIME types, that have magic pattern in the chemical-mime-data database.

The file cmd.magic.mime was created via XSLT conversion of the original chemical-mime-data database and can be used for KDE and file(1). One of the following package releases (probably 0.1.95) will provide it.

Update

Unfortunately this magic.mime file cannot be used for KDE. As pointed out by David Faure, KDE uses an older syntax that doesn't know e.g. the search type. So I have to find another solution for KDE or wait for better days.

Tuesday, February 13, 2007

Let file(1) recognize a chemical MIME type

Ok, it's 8 o'clock in the morning and I need some sleep. But first let's show you some small example, how to detect chemical MIME types with the file(1) command. Put the following stuff into your /etc/magic (the local magic data configuration file for the file(1) command):

0               string          VjCD0100                CDX binary file
>8              belong          0x04030201
>>12            bequad          0x0000000000000000
>>>20           beshort         0x0000
>>>>34          string          ChemDraw                written with ChemDraw
>>>>>42         string          x                       b%.4s
>>>>22          beshort         0x0080                  (new format)
>>>>22          beshort         0x0000                  (old format)

Now search for a CDX (ChemDraw binary) file and run the file command:

$ find . -name "*.cdx" -exec file "{}" ";"
./example.cdx: data (that's really broken)
./dummy.cdx: lif file (that's a CACTVS file in reality)
./x-chemdraw/structures25-27.cdx: CDX binary file written with ChemDraw 4.5 (old format)
./x-chemdraw/structures96-101.cdx: CDX binary file written with ChemDraw 4.5 (old format)
./x-chemdraw/structures40-48.cdx: CDX binary file written with ChemDraw 4.5 (old format)
./x-chemdraw/dimethylamine.cdx: CDX binary file written with ChemDraw 7.0 (old format)
./x-chemdraw/dimethylaminesimple.cdx: CDX binary file (old format)
./x-chemdraw/untitled.cdx: CDX binary file written with ChemDraw 8.0 (new format)
./x-chemdraw/structures1-12.cdx: CDX binary file written with ChemDraw 4.5 (old format)

Why I do this? The chemical-mime-data package contains magic pattern, that can be used to automatically create the rules for the file command too, so file(1) can determine the chemical MIME type too. Expect a stylesheet to extract this information from the database soon.

And now I will get some sleep.

Update

And here, how to recognize the MIME type. Add the following to /etc/magic.mime:

0               string          VjCD0100
>8              belong          0x04030201
>>12            bequad          0x0000000000000000
>>>20           beshort         0x0000
>>>>22          beshort         0x0080                  chemical/x-cdx
>>>>22          beshort         0x0000                  chemical/x-cdx

and run the file(1) command with the -i switch:

$ find . -name "*.cdx" -exec file -i "{}" ";"
./example.cdx: application/octet-stream
./dummy.cdx: application/octet-stream
./x-chemdraw/structures25-27.cdx: chemical/x-cdx
./x-chemdraw/structures96-101.cdx: chemical/x-cdx
./x-chemdraw/structures40-48.cdx: chemical/x-cdx
./x-chemdraw/dimethylamine.cdx: chemical/x-cdx
./x-chemdraw/dimethylaminesimple.cdx: chemical/x-cdx
./x-chemdraw/untitled.cdx: chemical/x-cdx
./x-chemdraw/structures1-12.cdx: chemical/x-cdx

And now I really get some sleep. Cheerio!