calibre

Bug #986044
Comment #28

Comment 28 for bug 986044

Revision history for this message

blorks (zemisterwhite) wrote on 2012-05-31:

#28

I just tested the new version of the driver with 0.8.53.

What I did :
- formatted SD from T1
- started calibre *without* the patch in order to create fucked up collections : got empty ones
- still *without* patch, plugged and added a new book : got a "removing..." entry in a collection
- started calibre *with* the patch , let in refresh metadata in the T1 : collections contain every expected book, but "removing..." entry was still there.
- plugged the T1 once again, removed a book on the SD, and added it again : got two "removing..." entries this time in the collection.

- removed all SD card collections (from T1)
- plugged the T1, let the metadata refresh : got empty collections
- plugged the T1 once again, added a new book : got a "removing..." entry in a collection

- formatted SD card from T1.
- plugged T1 and added books : all collections were fine
- plugged T1 added and removed books : all collections were fine
- removed all SD collections (from within T1)
- plugged T1, added a book : all collections were fine
- *BUT* database inspection shows that all id are low ones...

Conclusions :
1/ Repair feature did not work with my scenario.
I added some print statements and saw that "sequence_dirty" flag is not raised : if I understood the code, if books.id is "broken", then all tables will be repaired. But as previous analysis shown, id were all ok in database (for books, collection and collections), so repair procedure is not triggered.
A remark : I believe in the repair procedure, but the "sequence_dirty" flag has to be raised when needed (not only *.id, but also with collections.content_id and collections.collection_id).

2/ Collection management still broken with existing database.

3/ Starting from a fresh database seems to work BUT all ids are low ones, I don't know which side effects it will cause. I thought that there would be a mixup with internal/SD books when opening : my tests showed that it does not seem to happen : I opened book with "conflicting" ids from collections and from book list, no mixup happened.

So even there is no success, thank you for your work, we see that there is progress and hope you will not give up.

PS : I hope another tester confirms that, I don't want to be the only one to test... Setting up environment is easy once you guess how to deflate this xz file :)

I just tested the new version of the driver with 0.8.53.

What I did :
- formatted SD from T1
- started calibre *without* the patch in order to create fucked up collections : got empty ones
- still *without* patch, plugged and added a new book : got a "removing..." entry in a collection
- started calibre *with* the patch , let in refresh metadata in the T1 : collections contain every expected book, but "removing..." entry was still there.
- plugged the T1 once again, removed a book on the SD, and added it again :  got two "removing..." entries this time in the collection.

2/ Collection management still broken with existing database.

So even there is no success, thank you for your work, we see that there is progress and hope you will not give up.

PS : I hope another tester confirms that, I don't want to be the only one to test... Setting up environment is easy once you guess how to deflate this xz file :)