> > If I understand the msequencer code correctly:
> > Under '/Sequencer/State/Path' the path can be defined from where the msequencer gets the files, generates the xml, etc.
> > However, the new javascript code reads/writes the files to '<exp>/userfiles/sequencer/'
> >
> > If the path in the ODB is different to '<exp>/userfiles/sequencer/', it leads to quite some unexpected behavior. If '<exp>/userfiles/sequencer/' is the place where things should go, the ODB entry of the msequencer and the internal handling should probably adopted, no?
>
> Indeed there is a change in philosophy. Previously, /Sequencer/State/Path could point anywhere in the file system. This was considered a security problem, since one could access system files under /etc for example via the midas interface. When the new file API was
> introduced recently, it has therefor been decided that all files accessible remotely should reside under <exp>/userfiles. If an experiment needs some files outside of that directory, the experiment could define some symbolic link, but that's then in the responsibility of
> the experiment.
>
> To resolve now the issue between the sequencer path and the userfiles, we have different options, and I would like to get some feedback from the community, since *all experiments* have to do that change.
>
> 1) Leave thins as they are, but explain that everybody should modify /Sequencer/Stat/Path to some subdirectory of <exp>/userfiles/sequencer
>
> 2) Drop /Sequencer/State/Path completely and "hard-wire" it to <exp>/usefiles/sequencer
>
> 3) Make /Sequencer/State/Path relative to <exp>/userfiles. Like if /Sequencer/State/Path=test would then result to a final directory <exp>/userfiles/sequencer/test
>
> I'm kind of tempted to go with 3), since this allows the experiment to define different subdirectories under <exp>/userfiles/sequencer/... depending on the situation of the experiment.
>
> Best,
> Stefan
For me the option 3) seems the most coherent one.
Andreas |