search

karlb / wikdict-web

Web front end for WikDict dictionaries
https://www.wikdict.com
MIT License
16 stars 2 forks source link

[run_dev.sh] Resolve pysqlite3 leftovers #10

Closed hartwork closed 3 years ago

hartwork commented 3 years ago

After #3 I noticed that pysqlite3 does not even seem to be used any more (but Python's stdlib integration)…

karlb commented 3 years ago

The stdlib's pysqlite3 has no ICU support, unfortunately. This prevents good case insensitive searches, comparisons and similar operations for some languages. Skipping this can be ok for development on the web UI, but will cause worse query results in some cases.

I did not find a better way to install an ICU-enables sqlite3 python lib than the command removed in this PR.

hartwork commented 3 years ago

Is pysqlite3 used at the moment? From inspecting the virtualenv my impression was that it would take import pysqlite3 or from pysqlite3 import ... for pysqlite3 to be used, which I cannot find anywhere. My impression is: python stdlib is what's actually used at the moment. No? Please enlighten me :smiley:

karlb commented 3 years ago

Is pysqlite3 used at the moment? From inspecting the virtualenv my impression was that it would take import pysqlite3 or from pysqlite3 import ... for pysqlite3 to be used, which I cannot find anywhere. My impression is: python stdlib is what's actually used at the moment. No? Please enlighten me smiley

Hmm, it used to get imported under the sqlite3 name, but I can confirm that this does not happen anymore for me either. I'm not sure why and I'll have to dig deeper.

hartwork commented 3 years ago

Sounds good. Should I make CI PR #11 work for compiling pysqlite3 in the mean time (by installing the needed dependencies) or treat PR #10 as a blocker to PR #11 for now?

karlb commented 3 years ago

Since ICU apparently it is not working anyway, we can remove the custom pysqlite3 until that is fixed. Either via this PR or via #11.

hartwork commented 3 years ago

@karlb let's merge #10 then and I'll rebase #11 onto latest master after to get the bits from #10 out of it. I think that's the cleanest. Go?

karlb commented 3 years ago

@karlb let's merge #10 then and I'll rebase #11 onto latest master after to get the bits from #10 out of it. I think that's the cleanest. Go?

Go!