Has been removed to MyProgram 20100608
https://wiki.heprc.uvic.ca/twiki/bin/view/Main/CondorSOAPClient
The disadvantage of web condor (birdbath):
1) it can not produce new files by application itself except the condor output file (consle content).
2) it has no grain authorization control. All the web job is in the SPool, and all the user has the same power to the SPool.
It shows useless to solve the issues above by operating birdbath(client side). We have to do something in the server side.
Most of condor web files are located in condor_schedd.v6. Most of the apis are in the soap_scheddStub.cpp file.
Solution:
1) After a new job submitted, condor soap creates a directory using the cluster Id and proc id to it in the Spool dir. The cluster ID is generated by function NewCluster in src/condor_c++_util/Qmgmt.cpp;
2) if you want to add some attribute, such as email user, see ClassAd.cpp
3) You can transfer a executable as inputfiles, it will be transftered automatically to CN. You needn't worry whether Scheduler has the file. However, if you set cmd as an absolute file path, scheduler seems need the file in the path, cn needn't.