Following the collapse of the SPIDER work on a common SRT, it was decided to reconsider our choice of tools for configuration management independently of the other experiments.
A requirements document was prepared by Steve Fisher/RAL and Maya Stavrianakou/CERN and was reviewed and corrected largely according to the procedures of the old ASP.
Since then we have been looking for tools to go some way to meeting these requirements. People were invited to consider:
For each requirement, do you meet it fully, partially or not at all. If fully or partially - how is it satisfied? If partially or not - how much effort is involved to satisfy it and are you prepared to do it?
Assuming the answers encourage us to go further with your solution, which we expect to be the case, we will need to see a representative sample of the software currently maintained by SRT in the ATLAS Offline and Online converted to the new system. This sample will have to be big enough to convince us that we will not have scaling problems and will allow a final decision to be made.
What effort do you estimate for this task and is it possible for your new scheme to co-exist with the current SRT for the trial and for the change over period.
So far we have received only a few documents suggesting solutions as listed below.
A document on CMT from Christian Arnault which appears to go a long way towards meeting the requirements.
Tony Wildish has noted that rpm (the RedHat Package Manager) will meet many of our requirements and is well supported. He has written a note and has responded to a mail of mine.
An e-mail from Lassi A. Tuura which refers to an annotated copy of the requirements document indicates that the current ATLAS SRT meets many of our requirements and explains how it could be modified to meet others.
We have no formal document on this CMS product.
A document has been prepared which compares the known candidates. It is hoped to follow a procedure and time scale as shown in the plan. Previous versions of the plan may also be seen.