On 05/12/2010 12:50 PM, Iform wrote:
On 05/11/2010 08:20 AM, Olivier Tilloy wrote:
Hello,
Your description of the issue suggests that it may very well be an
issue
in pyexiv2 itself.
It would help a lot the investigation if you could let us know what
version of pyexiv2 you are using, and provide your script or a minimal
script that reproduces this behaviour.
Note that pyexiv2 0.2.0 had a nasty memory leak that was fixed in
pyexiv2 0.2.1, could that be the same issue?
Cheers,
Olivier
On 2010-05-11, S. J.<[email protected]> wrote:
Hi,
I have written a small python app that uses the pyexiv2 library.
However, when running the app, I see an interesting behavior.
My app walks through a directory and its subfolders, renaming each
file
according to it exif data metadata. It also writes the old
filename to
the xmp metadata fields as keywords. The at first the app functions
normally, memory usage is under control, the processor barely being
used
and near constant file I/O. However, it came to a point where memory
usage climbs, processor is maxed at 50% (only using one core). and
disk
I/O is very sporadic.
I get error messages to the console such as the below....
Warning: Directory NikonPreview, entry 0x0201: Data area exceeds data
buffer, ignoring it.
I am wondering what might be causing this. I cannot reproduce the
error
or the conditions using the exiv2 binary for windows. It seems to
read
all the metadata without errors. So I am not sure whether this is
a bug
in exiv2 or pyexiv2.
-Please Help
iform
Hi,
After running a few more tests...it looks like the culprit is not
reading the metadata, but writing it.
Included is a sample script that has the writeTags function of my app.
Behavior is observed using pyexiv2 .21
Hope this helps
iform
_______________________________________________
Mailing list: https://launchpad.net/~pyexiv2-developers
Post to : [email protected]
Unsubscribe : https://launchpad.net/~pyexiv2-developers
More help : https://help.launchpad.net/ListHelp
I did some more investigating and discovered that there is a known
performance issue with exiv2 .19 hanging on nikon jpgs. The exiv2 svn
fixed this bug (with .19 takes about 20secs to write a tag vs .1s with
svn-exiv2). I am also hoping that this resolved the "memory leak" in
pyexiv2.
The memory leak that I mentioned before only occurred when pyexiv2 was
processing problematic photos, which in turn happened to be all my nikon
dslr photos.
link to bug report
http://dev.exiv2.org/issues/show/677
hope this helps as well
iform
ps. thank you for the wonderful python binding.
_______________________________________________
Mailing list: https://launchpad.net/~pyexiv2-developers
Post to : [email protected]
Unsubscribe : https://launchpad.net/~pyexiv2-developers
More help : https://help.launchpad.net/ListHelp
