Search restriction bug when device is plugged in

[Manichean]

As of 0.7.18, I keep getting a weird bug when my device is plugged in. I haven't exactly been able to reproduce the bug, but it crops up now and then, seemingly when the library or device view were updated. The error message says:

ERROR: FEHLER: Unbehandelte Ausnahme: <b>TypeError</b>:search() got an unexpected keyword argument 'ignore_search_restriction'

Traceback (most recent call last):
File "site-packages\calibre\gui2\__init__.py", line 296, in dispatch
File "site-packages\calibre\gui2\device.py", line 759, in metadata_downloaded
File "site-packages\calibre\gui2\device.py", line 769, in refresh_ondevice_info
File "site-packages\calibre\gui2\library\views.py", line 205, in set_device_connected
File "site-packages\calibre\gui2\library\models.py", line 124, in set_device_connected
File "site-packages\calibre\gui2\library\models.py", line 271, in refresh
File "site-packages\calibre\library\caches.py", line 552, in refresh
TypeError: search() got an unexpected keyword argument 'ignore_search_restriction'

I have set up two restrictions based on a boolean custom field (searching for value "yes" and "no", respectively), the error occurs whether or not I try to search in the search bar or not.

Is anybody else getting this?

EDIT to add: XP SP3, Calibre 0.7.18 (german translation, should that matter), Cybook Gen3

[chaley]

Everyone will get it. The bug is triggered by plugging in a device when a restriction is in place. It can also arrive if a restriction is in place when books are set to or deleted from a device.

The bug has been there for weeks. The on-device changes just trigger it more often.

I have submitted a fix to Kovid for inclusion in the next release.

[Manichean]

Oh, okay. I didn't see any threads about it, and, since it only occured for me since the last update, I assumed it was new.

[chaley]

Originally Posted by Manichean

Oh, okay. I didn't see any threads about it, and, since it only occured for me since the last update, I assumed it was new.

There weren't any. Yours was the first.

Apologies for being unclear. Seems that today is my day for that. What I should have said was that the *flaw* has been there for weeks. Your exemplary report led me right to it. My guess is that until .18, the triggering conditions were so rare that either it didn't happen or people didn't bother to bring it up because they couldn't repeat it.

[Manichean]

Originally Posted by chaley

There weren't any. Yours was the first.

Apologies for being unclear. Seems that today is my day for that.

None needed.

Originally Posted by chaley

What I should have said was that the *flaw* has been there for weeks. Your exemplary report led me right to it. My guess is that until .18, the triggering conditions were so rare that either it didn't happen or people didn't bother to bring it up because they couldn't repeat it.

So, out of curiosity, did anything significant change? I didn't notice anything in the UI or the changelog, but that doesn't have to mean anything.

[chaley]

Originally Posted by Manichean

So, out of curiosity, did anything significant change? I didn't notice anything in the UI or the changelog, but that doesn't have to mean anything.

Sort-of depends on what you mean by significant.

The problem I fixed related to on-device matching. The problem was that a book on the device could match multiple books in the library, which makes no sense. Kovid and I worked out a scheme where the books will match the one-closest book in the library. This is good.

While testing this fix, I discovered that adding and deleting books in the library and on the device did not update on-device correctly. In some cases it was totally wrong. I attempted to fix that. Unfortunately, I broke some other things. That is the way of software -- changing one character 'here' can completely break what happens 'there'. Sigh.

As it turns out, the actual problem was introduced a few releases ago as part of redoing parts of the content server. Seems that no one noticed. Not surprising, given the corners of calibre I was playing in. My problem was assuming that those changes worked *there*, so I could reproduce them *here*. The code-daemons answered quickly -- "no, you can't".

(It is late and I am rambling. )

[Manichean]

Originally Posted by chaley

Sort-of depends on what you mean by significant.

The problem I fixed related to on-device matching. The problem was that a book on the device could match multiple books in the library, which makes no sense. Kovid and I worked out a scheme where the books will match the one-closest book in the library. This is good.

While testing this fix, I discovered that adding and deleting books in the library and on the device did not update on-device correctly. In some cases it was totally wrong. I attempted to fix that. Unfortunately, I broke some other things. That is the way of software -- changing one character 'here' can completely break what happens 'there'. Sigh.

As it turns out, the actual problem was introduced a few releases ago as part of redoing parts of the content server. Seems that no one noticed. Not surprising, given the corners of calibre I was playing in. My problem was assuming that those changes worked *there*, so I could reproduce them *here*. The code-daemons answered quickly -- "no, you can't".

(It is late and I am rambling. )

Thanks for the explanation. I actually know that "process" of accidentally breaking things that cannot possibly be broken by the code you're changing right now... Though far from being a programmer, I've written some code myself, and, at the moment, I'm trying to learn Python. I really need to take a weekend off and just absorb the whole tutorial in one go, I think...

It's late here as well, can you tell?