Page MenuHomeHEPForge
Files F8310459

TODO
No OneTemporary
Actions

TODO
View Options

Rivet validator TODOs
=====================
SUBMISSION
* Make it possible to add more jobs to a given run ID
* Provide short commands for shell
list == l,ls, monitor == m,mon,cat...
Does cmd have a nice way to define command synonyms without also duplicating
the interactive docs?
* Remove underscore on run output dirs
This underscore was mainly helpful for globbing when deleting */*/_* before.
Now that all output for one run ID is under one run output dir, is it useful?
* Standardise status code symbols
enum { QUEUED, RUNNING, FINISHED, FAILED } -> dict somewhere?
* bash completion for gen list
* Multi-runs in one analysis
Will be needed... previously we might use 5 runs for one analysis, of which 2
would also be used for one othe analysis. This degree of flexibility may be
unobtainable without making the submitter so general that it makes nothing
easier!
* Use enhancement function in sampling for Sherpa, Hw++
For analyses which need stats in high pT bins, use weighted analyses and an
enhancement function (e.g. pT**2) in the generation. We can get away without
weighted events if using profile histos, but this is nicer/more
general. Rivet should already handle the weights. Also possible for Pythia 6?
Other gens?
* Add prerun, postrun scripts
Use them if they are there (again, fall-back from analysis dir to gen dir).
Prerun could be used for e.g. setting up FIFO, post-run for merging several
runs contributing to one analysis.
* Allow on the fly param customisation
cf. Professor random param changes atop a standard base tune
* Grid distribution support
Just running glite commands to start
* Make submission/validation a bit more like an installed program
Use setup.py mechanism to install Py modules and scripts, rename run ->
mcvalidator-run (for example)
* Provide a "bootstrap validation tree" command
Just copy across the default area to a new location.
* Grid pilot jobs with DAG
* More advanced templating support
Use Cheetah for templating, with a set of standard template tokens,
cf. prof-rungen. This may not be needed until we want to use the submission
system as a platform for Professor scans.
OUTPUT
* Find reference histogram automatically for each histogram
(per analysis doesn't work, because multiple analyses might be in one
hist.aida)
* Fill legends in plots
* Perform run merging
* Make Web pages nicer: two-column, maybe expand-plot on click
* Create validation matrix.

File Metadata

Mime Type
text/plain
Expires
Sat, Dec 21, 6:33 PM (1 h, 34 m)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
4023363
Default Alt Text
TODO (2 KB)

Event Timeline