Re: AW: [geclipse-dev] Minutes of the JSDL meeting from today

[Ariel Garcia <garcia@xxxxxxxxxx>]

Hi Kasia, all,

i agree that the "input/output" type of file could be left for later 
implementation if required, from what i understood it is an "edit the 
file"  usecase, getting a the output with a different filename shouldn't 
be a big problem.

In any case i wanted to make a small comment on another point:

at least in glite's jdl (i don't know in jsdl)  you can specify two 
different set of i/o files:
 - the i/o sandbox which are files transfered to/from your 
desktop/user-interface node from/to the place where the job runs
- the OutputData which are job-output files which are uploaded to some SE 
after the job is done.
(there is also InputData but that is silly... those values are used by the 
RB/WMS for resource matching but the files are not automatically put where 
your job runs, you have to download the files 'by hand' from your job)

Regaring the Sandbox the syntaxis allows only for one filename, so the file 
does not get renamed and it is copied to the folder where your job gets 
run (ie, no subfolders)

On the contrary, OutputData allows for src AND dst names, 
  OutputData = {[
   OutputFile="temp.txt";
   LogicalFileName="lfn:/grid/dgtest/anotherTest55";
  ]};

Is that possibility allowed in jsdl? if yes it would be nice to foresee 
that in the jsdl editor.


Regarding look&feel of the wizard, perhaps an idea would be to put an icon 
with a left or right arrow besides each input or output line on the table, 
to help the user see which kind of transfer it is.

my 2cts :)
Cheers,
Ariel