On Tue, Nov 8, 2011 at 1:23 AM, Ross Walker <ross.rosswalker.co.uk> wrote:
>
> One question. How does the script handle deleting files, creating blank
> directories and or brand new files? - I think several versions of patch do
> not support patching a non-existant file for example so you need to be
> careful here.
>
The ONE detail I didn't touch on in my lengthy document. Since it parses
file names to determine if we patch them or not, it knows if it should
create a blank file for any file that doesn't exist (first it creates the
containing directories if they're not present). Same as Unix "touch". For
deleted files, I don't know how to parse that out of the patch, so for now
it just lets patch delete the files if it's able to (according to the
manpage I saw, POSIX standard declares that empty files are deleted, as
well as any directories that are emptied as a result.
If any of the Amber patches in our repo right now (outside of 17) do that,
can you check to see if my script handles it correctly, possibly on
different platforms? If any patches do, it'll definitely be one of the
CUDA feature patches, I just don't know about it.
Thanks!
Jason
> All the best
> Ross
>
> > -----Original Message-----
> > From: Jason Swails [mailto:jason.swails.gmail.com]
> > Sent: Monday, November 07, 2011 10:13 PM
> > To: AMBER Developers Mailing List
> > Subject: [AMBER-Developers] bugfix update script
> >
> > Hello everyone,
> >
> > I've put together a prototype script for managing bug fixes for future
> > versions of Amber and AmberTools. This should make it much easier for
> > the
> > user to apply the latest bug fixes without having to jump through the
> > hoops
> > they used to. I think the benefit this provides is worth the fact that
> > it's written in Python :-P (it works on Python 2.4 -- Python 2.7, so it
> > should work on every system we've ever tested it on). It's in the
> > master
> > branch (but if you try to use it there it will just try applying
> > Amber11
> > and AmberTools 1.5 patches to your git repo -- don't do that :)
> >
> > To keep this email short, I've attached 3 documents describing how to
> > use
> > this script, how the script was implemented (which may elucidate
> > weaknesses), as well as the format restrictions this places on bug
> > fixes
> > (basically, just let git do it, but it should not be too different than
> > what we have now, except for people not used to making individual
> > patches
> > that need to be applied from AMBERHOME).
> >
> > As I think this will do a lot to improve user-friendliness, I would
> > *greatly* appreciate any kind of feedback and discussion regarding this
> > script. If you can spare it, please take some time, test it out on a
> > fresh
> > AmberTools 1.5/Amber11 tree, and tell me what you think. How can it be
> > improved? Are there any problems with it? Anything unclear? As
> > evidenced
> > by AT15_Amber11.py (which didn't get much if any testing/suggestions
> > before
> > it was thrown into production), this will greatly benefit from more
> > than
> > just my ideas and perspective.
> >
> > Thanks in advance!
> > Jason
> >
> > P.S. -- A tip from when I was testing this script:
> >
> > To aid in my own testing, I created a new git repository from the
> > AmberTools 1.5/Amber11 checkout with the patch_amber script in it. A
> > quick
> >
> > git init; git add .; git commit -m 'initial commit'
> >
> > command in that fresh directory (after you copy in patch_amber.py) will
> > create the git repository and initialize it. Then, you can always
> > restart
> > from the beginning with
> >
> > git clean -f -x -d; git checkout .
> >
> > from the amber11 directory (that's the only way it'll clean the whole
> > directory).
> >
> > Some things I already know and are not bugs -- Amber bugfixes 9 and 12
> > will
> > fail unless you use --ignore-fails, since both of those bug fixes try
> > to
> > fix configure from AmberTools 1.4. The script will only ignore failed
> > HUNKs if you force it to (this is intentional). Bugfix.17 for Amber
> > will
> > complain that no files are there to fix. That's because the patch is
> > just
> > the header, the actual patch is a hacked-up, bzipped tarball of the
> > likes
> > that we won't see again (since this new script will be able to allow us
> > to
> > handle that patch done "correctly").
> >
> > --
> > Jason M. Swails
> > Quantum Theory Project,
> > University of Florida
> > Ph.D. Candidate
> > 352-392-4032
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Mon Nov 07 2011 - 23:00:03 PST
