After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 324425 - Clear existing database of Invalid size of entry
Clear existing database of Invalid size of entry
Status: RESOLVED FIXED
Product: f-spot
Classification: Other
Component: Browsing
0.1.x
Other All
: Normal minor
: ---
Assigned To: F-spot maintainers
F-spot maintainers
: 326999 (view as bug list)
Depends on:
Blocks:
 
 
Reported: 2005-12-18 22:36 UTC by Bengt Thuree
Modified: 2007-05-24 20:02 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
Picture with weird UserComment (487.62 KB, image/jpeg)
2006-02-26 02:07 UTC, Bengt Thuree
Details
Example photo (754.04 KB, image/jpeg)
2006-08-10 12:08 UTC, Arif Lukito
Details

Description Bengt Thuree 2005-12-18 22:36:53 UTC
Please describe the problem:
I have not set the titel on my pictures yet, but I found the following
text there.
 
Invalid size of entry (8, expected 0 * 1).
 
I did set the titel (in picture view, just below the picture), and this
message/titel dissapeared and the one I set is there instead. As it
should.

Steps to reproduce:
1. Import a new photo (with no tags nor title)
2. Go to Photo view mode (not browser mode)
3. Check the title


Actual results:
Invalid size of entry (8, expected 0 * 1).

Expected results:
A blank title (or "Please enter a subject/title for this picture" message)

Does this happen every time?
Yes

Other information:
This has been confirmed for Canon S70, Nikon D1, and Canon 20D.
Larry commented : Sounds like libexif if complaining an returning that text as
the result. I really need to stop using libexif and switch to the code in f-spot.
Comment 1 Pasi Savolainen 2005-12-18 23:09:52 UTC
Image behind this url shows same behaviour (1.1MB):
http://varg.dyndns.org/psi/random/dsc_1378.jpg
Comment 2 Pasi Savolainen 2005-12-19 01:13:39 UTC
My libexif is Ubuntu/dapper, package libexif12 version 0.6.12-2.
Comment 3 Arif Lukito 2005-12-19 06:43:44 UTC
Distro: Gentoo
libexif ver: 0.6.12
Comment 4 Bengt Thuree 2005-12-19 12:33:54 UTC
Ubuntu/Breezy, package libexif12 version 0.6.12-2
Comment 5 Michael R Head 2005-12-29 21:24:48 UTC
Here's another from my Canon A520
http://www.core.binghamton.edu/~burner/StomachButt/img_2686.html
Comment 6 Bengt Thuree 2006-01-16 04:45:35 UTC
A duplicate bug of this one was filed (bug #326999)
Comment 7 Gabriel Burt 2006-01-17 06:02:36 UTC
*** Bug 326999 has been marked as a duplicate of this bug. ***
Comment 8 Jamie Wilkinson 2006-01-26 23:03:17 UTC
Debian unstable, libexif 0.6.12-2
Comment 9 Bengt Thuree 2006-02-20 05:14:55 UTC
Any news on this one?
Comment 10 Jamie Wilkinson 2006-02-20 05:22:37 UTC
Actually I haven't seen this behaviour recently; I'm still using a relatively old CVS checkout of f-spot with my own modifications and libexif 0.6.12-2 from Debian unstable, and my last set of imports over the last fortnight didn't exhibit this bug at all.
Comment 11 Bengt Thuree 2006-02-20 10:05:05 UTC
sorry, it is still there for me.
Same libexif (Dapper/Ubuntu), latest CVS checkout though.
Comment 12 Larry Ewing 2006-02-25 21:56:48 UTC
this should be fixed now.
Comment 13 Bengt Thuree 2006-02-26 01:23:27 UTC
Seems to be something within Exif.
Did some extra tracing... :)
Printout from Exif.cs, Lookup function
public ExifEntry Lookup (Tag tag)
{
 Assemble ();
 foreach (ExifEntry entry in entries) 
  System.Console.WriteLine ("entries >>{0}<< - >>{1}<<", entry, entry.Value);


entries >>Exif.ExifEntry<< - >>1/4 sec.<<
entries >>Exif.ExifEntry<< - >>f/2.8<<
entries >>Exif.ExifEntry<< - >>Exif Version 2.1<<
entries >>Exif.ExifEntry<< - >>2000:08:29 21:56:29<<
entries >>Exif.ExifEntry<< - >>2000:08:29 21:56:29<<
entries >>Exif.ExifEntry<< - >>Y Cb Cr -<<
entries >>Exif.ExifEntry<< - >>3.00<<
entries >>Exif.ExifEntry<< - >>131072/65536 sec. (APEX: 2)<<
entries >>Exif.ExifEntry<< - >>f/2.8<<
entries >>Exif.ExifEntry<< - >>0.0<<
entries >>Exif.ExifEntry<< - >>2.97<<
entries >>Exif.ExifEntry<< - >>65.5 m<<
entries >>Exif.ExifEntry<< - >>Center-Weighted Average<<
entries >>Exif.ExifEntry<< - >>Flash fired.<<
entries >>Exif.ExifEntry<< - >>5.4 mm<<
entries >>Exif.ExifEntry<< - >>310 bytes unknown data<<
entries >>Exif.ExifEntry<< - >>Invalid size of entry (8, expected 0 x 1).<<
entries >>Exif.ExifEntry<< - >>FlashPix Version 1.0<<
entries >>Exif.ExifEntry<< - >>sRGB<<
entries >>Exif.ExifEntry<< - >>1600<<
entries >>Exif.ExifEntry<< - >>1200<<
entries >>Exif.ExifEntry<< - >>7766.99<<
entries >>Exif.ExifEntry<< - >>7741.94<<
entries >>Exif.ExifEntry<< - >>Inch<<
entries >>Exif.ExifEntry<< - >>One-chip color area sensor<<
entries >>Exif.ExifEntry<< - >>DSC<<
Comment 14 Bengt Thuree 2006-02-26 02:07:03 UTC
Created attachment 60139 [details]
Picture with weird UserComment

Larry, I used Eye of Gnome and checked some of my photos, and I can see this User Comment here as well? But I do not see this when I use a Windows program (IrfanView).

I attach the photo I have been playing with.

Could this be related to libexif? and might be fixed if we use our own?
Comment 15 Bengt Thuree 2006-02-26 02:49:31 UTC
Also, we need an Update function to remove a possible "Invalid size of entry *" comment, both in database, as well as in exif.
Comment 16 Bengt Thuree 2006-05-15 15:32:42 UTC
Any news on this one?
Comment 17 Stephane Delcroix 2006-06-14 12:50:26 UTC
I did not see the problem anymore, since Larry fixed it (comment #12).

Bengt, it's possible that you can still see it with eog because f-spot writes data in the Metadata of the file (if the preference is checked).
You can remove that from the db using an sql command(do you want it ? really ?), and the new (empty) comment will be written next time you do something on an image (tagging, changing time, ...)

Since the bug is fixed, I guess we can close it. Btw, it could be a good help to users if we add a small 'upgrade database' script with the next release, removing all the bad comments and incrementing the db version.
Comment 18 Bengt Thuree 2006-06-14 13:08:30 UTC
ok, I close this one.
Comment 19 Bengt Thuree 2006-06-14 22:59:01 UTC
I re-open and change the title of this one, to reflect the fact that we need to update the existing databases, and remove this title.

Invalid size of entry (8, expected 0 * 1).

I think it is enough to check for

Invalid size of entry
and if this is found, clear the title.
Comment 20 Arif Lukito 2006-08-08 05:17:33 UTC
I'm still seeing this on metadata browser
User Comment	Invalid size of entry (8, expected 0 x 1).
Comment 21 Bengt Thuree 2006-08-10 11:25:22 UTC
Could you attach a photo that you get this problem with when you import it to F-Spot?
Comment 22 Arif Lukito 2006-08-10 12:08:32 UTC
Created attachment 70629 [details]
Example photo
Comment 23 Bengt Thuree 2006-08-10 12:23:29 UTC
Just tried to import this photo in CVS version of F-Spot, and can confirm that this photo show's the "Invalid size of entry (8, expected 0 x 1)." in Metadata browser.


I also imported it with my XMP Import patch, and got the following result.
1 Subject = >><<
2 Predicate = >>http://ns.adobe.com/exif/1.0/UserComment<<
3 Object = >>_<<
4 Meta = >>_<<
5 Uri = >><<
6 Title = >>UserComment<<
7 Value = >>null<< --- Collection YES
8 type = >>Alt<<
9 title = >>22-rdf-syntax-ns#li<<
10 value = >><<

UserComment is "" in this case.
The XMP Import patch, removes all illegal characters from the start/end of the string though
Comment 24 Bengt Thuree 2006-09-17 23:26:14 UTC
Could this one be related to bug #356259
(The character's Bug #356259 refers to is removed in the patch I used in comment #23)
Comment 25 Stephane Delcroix 2007-01-29 15:41:20 UTC
I added a reminder note in the Updater.cs to consider cleaning this during the next major db upgrade
Comment 26 Stephane Delcroix 2007-03-06 18:08:08 UTC
I'm about to fix this in SVN, but I need a photos.db with those invalid entries.
So, if you have one, send it to me by email
Comment 27 Stephane Delcroix 2007-05-24 20:02:42 UTC
fixed in r3110