pyexiv2-developers team mailing list archive

[Damien Moore <damienlmoore@xxxxxxxxx>]

Hi Olivier,

>It fully supports reading from sidecar files, but apparently has a
>problem writing back the modified data.

well, reading is a good start. how about reading from one file (either
sidecar or images) and writing the data (perhaps with changes) to a new
sidecar file?

>This raises no error, however when inspecting the file afterwards, the
>value for Xmp.dc.creator is still 'John Doe'.

I guess it should be possible to test this with the exiv2 command line
tool(?)

>I'm not sure how to interpret that, it looks like a bug to me. Would
>that be a show-stopper for you?

I want to give the user the option to choose XMP or IPTC (or both to keep
certain fields in sync). If they choose XMP, ideally they will have the
option to modify the original image file or use sidecars.

cheers,
Damien

On Wed, Jan 27, 2010 at 10:36 AM, Olivier Tilloy <olivier@xxxxxxxxxx> wrote:

> Hey Damien,
>
> Damien Moore a écrit :
> > Hi Olivier
> >
> > [...]
> >
> >>(including the long-awaited XMP metadata)
> >
> > Another welcome change, but one I have a question about. Does the XMP
> > support just allow reading/writing that metadata from the image binary
> > or is there support for reading/writing sidecars as well?
>
> It fully supports reading from sidecar files, but apparently has a
> problem writing back the modified data. To be honest that's something I
> hadn't tested before, and I don't really know how exiv2 behaves in this
> regard.
> I've simply tested with the StaffPhotographer-Example.xmp file that
> comes with the test data in exiv2 trunk (can be downloaded at
>
> http://dev.exiv2.org/repositories/changes/exiv2/trunk/test/data/StaffPhotographer-Example.xmp
> ),
> here is what I've done:
>
> import pyexiv2
> m = pyexiv2.ImageMetadata('StaffPhotographer-Example.xmp')
> m.read()
> # At this point metadata is correctly read and accessible, e.g.:
> m['Xmp.dc.creator'].value # returns [u'John Doe']
> # Changing the value:
> m['Xmp.dc.creator'].value = [u'Clark Kent']
> # And writing back to the file
> m.write()
>
> This raises no error, however when inspecting the file afterwards, the
> value for Xmp.dc.creator is still 'John Doe'.
> I'm not sure how to interpret that, it looks like a bug to me. Would
> that be a show-stopper for you?
>
> > cheers,
> > Damien
>
> Cheers,
>
> Olivier
>