- Apr 12, 2018
-
-
Antoine Beaupré authored
-
- Mar 20, 2018
-
-
Antoine Beaupré authored
-
- Mar 13, 2018
-
-
Antoine Beaupré authored
-
Antoine Beaupré authored
I have tried reverse-engineering the Kobo database format, and it's a real mess. The `content` table, for example, has about 90 *columns*, most of which are basically useless. There are multiple rows per book, one per book but also one per chapter per book (instead of having, say, a separate `chapters`) table. There is a lot of user tracking junk in there as well, which tells me the reader is spying on its users and reporting that information back to headquarters, something which we already knew about, but it's kind of creepy to see in action. Even worse: those results are completely useless in trying to improve the current update mechanisms. Changes to the database are not picked up by the GUI at all, and changes in the GUI are not written in the database until a reboot or USB connect. This will make it hard, if not impossible to find a better way to update the contents than the current `fake-usb-connect` mechanism. Collections/shelves (#20) will also be difficult to implement, unless the DB is updated *while the USB dialog is open*, which is really not easy to do, because the database then gets unmounted! So this was a full day of work wasted, thanks to Kobo's failed engineering. It's a miracle this thing works at all, when you think about it. This makes [okreader] really interesting again... [okreader]: https://github.com/lgeek/okreader/
-
Antoine Beaupré authored
We had two GOARCH flags in the tarball target, and Make obviously considers only the latter, so they are redundant. While we're here, reorder and regroup those variables so they make more sense.
-
Antoine Beaupré authored
-
- Mar 06, 2018
-
-
Antoine Beaupré authored
-
Antoine Beaupré authored
lint needs to run after build or depends are missing. we simply skip subdirectories for now, as vendors are not clean
-
Antoine Beaupré authored
Use the binary releases because it's the recommended way to install it upstream. We can start using the Debian packages once Debian Buster is released and used in the official golang packages. We run lint earlier now because linting fails on some vendor/, in particular sqlite.
-
Antoine Beaupré authored
This currently means the latest and greatest 1.10
-
Antoine Beaupré authored
still need to change doAPI and factor out common code, but this is improving
-
Antoine Beaupré authored
-
- Feb 22, 2018
-
-
Antoine Beaupré authored
Trying to make sure we have a way to pin dependencies during development. This doesn't fix fundamental issues with go get: that's a broader problem with the go ecosystem that dep can't fix directly. We fix this with CI right now, it's a packaging problem. We also don't commit vendor/ as I object to the duplication of code. I assume that the presence of the .lock file provides sufficient protection against spurious upstream changes (e.g. with -vendor-only) that we don't need to keep a local copy.
-
- Dec 21, 2017
-
-
Antoine Beaupré authored
-
Antoine Beaupré authored
This will, hopefully, clarify how each setting can be specified in the configuration file. Closes: #23
-
- Nov 11, 2017
-
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
- Oct 17, 2017
-
-
Antoine Beaupré authored
add tag filtering & document Closes #19 See merge request !5
-
- Oct 16, 2017
-
-
Bogdan Cordier authored
- Only articles tagged with at least one tag defined in config are downloaded. - Tags must be a comma separated list of strings. - Filtering is not case sensitive. - Spaces before & after each tag are trimmed. Fix #19
-
- Oct 05, 2017
-
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
Antoine Beaupré authored
See: #22
-
- Jun 20, 2017
-
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
- Mar 07, 2017
-
-
Antoine Beaupré authored
i do not believe this is necessary now that we close the database properly see #6 for detailed discussion
-
Antoine Beaupré authored
15 seconds is just way too long for people to wait for the database to refresh see #6 for detailed discussion
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
Antoine Beaupré authored
See #14
-
Antoine Beaupré authored
Closes: #16
-
Antoine Beaupré authored
-
Antoine Beaupré authored
-
- Mar 06, 2017
-
-
Antoine Beaupré authored
-
- Mar 05, 2017
-
-
Antoine Beaupré authored
-