Ok, thanks for the hint! Maybe and advantage of keeping these strings in HDF5 is that if the dataset they refer to is deleted they also go to oblivion. Otherwise I have to keep things synchronized. Or decorate deletion of datasets. Or use the undo mechanism. Just trying to wrap my head around the possibilities of PyTables.
-á. On Thu, Mar 15, 2012 at 18:08, Anthony Scopatz <scop...@gmail.com> wrote: > Cool idea, but why not just have a log file on the side that the decorator > writes to? HDF5 only allocates a certain amount of space for attributes / > attribute names. (You can check the spec but I think it is something like 64 > k.) So if you are writing an excessive number of attributes you may run > into problems. If it is really important that this log goes into the HDF5 > file itself, I would consider looking at the variable length string atom for > VLArrays: http://pytables.github.com/usersguide/libref.html?highlight=vlstring#vlstringatom > > On Thu, Mar 15, 2012 at 8:03 AM, Alvaro Tejero Cantero <alv...@minin.es> > wrote: >> >> Hi, >> >> Here's my last question for today (I sent them separately because they >> are quite unrelated). >> >> I am thinking of writing a python decorator that for any processing >> function (e.g. band-pass filter of median of data[0:3,:]) logs to the >> attributes of the target HDF5 column >> >> * the name of the function, >> * the location of the repository where it lives and a string >> identifying the commit, >> * the arguments that were passed to it (or at least the parameters >> that tune the function; see below) >> >> The goal is to keep my data processing functions generic, yet be able >> (for scientific traceability) to track how a particular result was >> generated. I am still thinking how to report arguments that are long >> arrays without making all of the functions accept as an argument a >> HDF5 tree location string. >> >> Any thoughts / prior art? >> >> regards, >> >> Álvaro. >> >> >> ------------------------------------------------------------------------------ >> This SF email is sponsosred by: >> Try Windows Azure free for 90 days Click Here >> http://p.sf.net/sfu/sfd2d-msazure >> _______________________________________________ >> Pytables-users mailing list >> Pytables-users@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/pytables-users > > > > ------------------------------------------------------------------------------ > This SF email is sponsosred by: > Try Windows Azure free for 90 days Click Here > http://p.sf.net/sfu/sfd2d-msazure > _______________________________________________ > Pytables-users mailing list > Pytables-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/pytables-users > ------------------------------------------------------------------------------ This SF email is sponsosred by: Try Windows Azure free for 90 days Click Here http://p.sf.net/sfu/sfd2d-msazure _______________________________________________ Pytables-users mailing list Pytables-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/pytables-users
