MobileRead Forums - View Single Post - Testers wanted for 'series' custom columns

chaley · 06-20-2010, 04:43 PM

After discussion with Kovid, I have added a new type of custom column: series. I am looking for some testers.

Series custom columns are like the standard series column. They contain a series name that many books can refer to, and a 'n.nn' value that is individual to a book. Editing via edit metadata and directly on the library view is supported. The series name is searchable and appears in the tag browser. Like other custom fields, you cannot yet use them in collections, but that will come.

The series type was added to permit books to be in more than one series at a time. For example, Weber's Honorverse books have at least an Honorverse temporal order, a publishing order, and a sub-series order. Bradley's Darkover books have the same characteristics, with the added complexity that the sub-series are threaded together. Using the new column type, you can keep track of the ones that interest you.

Now to testing: I have tried it in all the permutations that I can think of, but hard experience has shown that I don't think of them all. My hope for opening it for testing is to get confidence that series columns work and that nothing else has been broken. Even just a few people trying it will raise confidence levels substantially.

Some things to test: I had to touch a little bit of code in a lot of places. I had to change both edit metadata functions (single and bulk), and there is some possibility that I broke them. I also changed the library view to both display and edit the series columns, changed sorting for the obvious reason, and modified the create custom column display. All of these should be tried by different fingers than mine. In addition, the version should be simply used as normal, because there might be unexpected behavior somewhere else.

To test, you must run from source. The source repository is lp:~cbhaley/calibre/charles_test/. If you have never run from source, see http://calibre-ebook.com/user_manual/develop.html. The process isn't overly difficult, but it does make a few demands that not everyone is prepared for. The most challenging is installing the version control environment 'bzr', used for downloading the source.

There are no changes to the database schema, so the risk of damaging a your database is extremely low. That said, as always it would be good to have a backup.

There is only one 0.7.4 compatibility problem. Once you create series columns, you will not be able to run 0.7.4 until you delete those columns. If you forget to delete them, 0.7.4 will complain bitterly about an unknown column type '#series' and not start properly. If you want to back up to 0.7.4, delete the columns using the test version, or by using 0.7.4's command line command 'calibredb remove_custom_column'. Alternatively, and better, would be to run it on a copy of your library.

Thank you in advance for your help.

06-20-2010, 04:43 PM	#1
chaley Grand Sorcerer Posts: 11,739 Karma: 6997045 Join Date: Jan 2010 Location: Notts, England Device: Kobo Libra 2	Testers wanted for 'series' custom columns After discussion with Kovid, I have added a new type of custom column: series. I am looking for some testers. Series custom columns are like the standard series column. They contain a series name that many books can refer to, and a 'n.nn' value that is individual to a book. Editing via edit metadata and directly on the library view is supported. The series name is searchable and appears in the tag browser. Like other custom fields, you cannot yet use them in collections, but that will come. The series type was added to permit books to be in more than one series at a time. For example, Weber's Honorverse books have at least an Honorverse temporal order, a publishing order, and a sub-series order. Bradley's Darkover books have the same characteristics, with the added complexity that the sub-series are threaded together. Using the new column type, you can keep track of the ones that interest you. Now to testing: I have tried it in all the permutations that I can think of, but hard experience has shown that I don't think of them all. My hope for opening it for testing is to get confidence that series columns work and that nothing else has been broken. Even just a few people trying it will raise confidence levels substantially. Some things to test: I had to touch a little bit of code in a lot of places. I had to change both edit metadata functions (single and bulk), and there is some possibility that I broke them. I also changed the library view to both display and edit the series columns, changed sorting for the obvious reason, and modified the create custom column display. All of these should be tried by different fingers than mine. In addition, the version should be simply used as normal, because there might be unexpected behavior somewhere else. To test, you must run from source. The source repository is lp:~cbhaley/calibre/charles_test/. If you have never run from source, see http://calibre-ebook.com/user_manual/develop.html. The process isn't overly difficult, but it does make a few demands that not everyone is prepared for. The most challenging is installing the version control environment 'bzr', used for downloading the source. There are no changes to the database schema, so the risk of damaging a your database is extremely low. That said, as always it would be good to have a backup. There is only one 0.7.4 compatibility problem. Once you create series columns, you will not be able to run 0.7.4 until you delete those columns. If you forget to delete them, 0.7.4 will complain bitterly about an unknown column type '#series' and not start properly. If you want to back up to 0.7.4, delete the columns using the test version, or by using 0.7.4's command line command 'calibredb remove_custom_column'. Alternatively, and better, would be to run it on a copy of your library. Thank you in advance for your help.