ATLAS-ALPHA Initial Status
This is a summary of comments I received regarding versions of ATLAS code that
collaborators have installed on their AXPs and their status.
Comments made by several people:
- Desire for standard multi-platform verification/testing before general
- I never know when a new release is being made!
- As a huge collaboration (larger than 90% of businesses in the USA)
we have to get away from the notion that the only people working on the
code are people resident at the lab. It's not just wrong, it's damaging!
- The code changes too often to bother to keep up with it...I hope the
user's group can combine to keep the code maintained.
Comments worthy of quotation (names withheld to protect the guilty).
- These HPs are gutless...why doesn't the code run on the super fast
- The code is sufficiently complex that it seems someone is required
to debug nearly full time. While this may be natural for a "running
experiment" it is a major overhead when groups are heavily committed to
preparing pieces of the ATLAS detector and are supporting analysis
software committments on other experiments.
- The lack of enthusiasm by the code authors to support "anything"
besides HPs makes it difficult to imagine that one will ever get to
the point of "actively debugging ATLAS software" on any platform other
than an HP - which is a shame.
- My experience working on OSF and HP-UX is that
something you read in a book is likely to work on OSF but never works on
HP. i.e OSF is much more 'standard' than HP.
- We need clear instructions and WORKING examples. We also need proper
old, pro new versions where pro actually works. The new can change as
much as it likes but the pro should only change say 2 to 4 times a year.
Ransom Stephens -