Canan Talayhan writes: > It seems after testing lots of pages this one escaped me since I only > tested the working case. > > Please find the quick fix in the link below. > https://pastebin.ubuntu.com/p/s7tWyPHZ8F/ Great, that fixes the issue with the revision comparison page. > I'm looking forward to making another contribution. Could you please > review it as soon as possible? I've gone ahead and merged this. I made some changes to the indentation, I've generally just left that to Emacs, so that's effectively the indentation style currently. I also tweaked the wording in the commit message. As for what to do next, it would be good to start looking at some stuff that's more related to the project topic. Part of the revision processing that I believe is quite slow and hopefully can be improved is populating the package_metadata table. The relevant lines in the job output look something like this: debug: Starting querying the temp_package_metadata debug: Finished querying the temp_package_metadata, took 1902 seconds That output comes from the with-time-logging bit around here [1]. It's a single query, generated by temp-table-select-query which is taking around 30 minutes it seems, and I'd hope either the query can be made faster, or some other faster way of doing what needs doing here can be found. 1: https://git.savannah.gnu.org/cgit/guix/data-service.git/tree/guix-data-service/model/utils.scm#n333 insert-missing-data-and-return-all-ids is used in a few places, but this specific issue arises when called from here [2]. 2: https://git.savannah.gnu.org/cgit/guix/data-service.git/tree/guix-data-service/model/package-metadata.scm#n373 Making inserting package metadata faster is the overall goal, but I'd suggest first just trying to reproduce the slow query outside of the revision processing process. That way you'll be able to look at what the query is doing and quickly test changes. The approach I'd recommend is, make yourself a realistic temp_package_metadata table by populating it with all the package_metadata entries for a single revision already in your local database. Then construct and try the slow query, and see how long it takes, and look at the query plan (run the query with EXPLAIN at the start). Do let me know if you have any questions or get stuck, I'll hopefully be around on IRC, and if I don't respond within a few minutes there, just email me.