Failing literal search

[mikesname]

Archives nationales d'outre-mer

Only works when quoted.

[juntezhang]

You can enable the elision filter to ignore article elisions. This requires a full reindex though for documents that have these tokens.

https://solr.apache.org/guide/solr/latest/indexing-guide/language-analysis.html#french

[mikesname]

@juntezhang thanks, I'll check that out! Good to hear from you, hope things are going well in NL.

[mikesname]

It turns out that this issue was a bit more straightforward than anticipated, since we already have fields set up with the elision filter for various languages. We just weren't using them in this search 🙍‍♂️.

This probably wouldn't have been noticed, except we added a "mimimum must match" parameter more recently, which in combination the language analysis setting for "generic" text, caused this search to fail.

I have put a fix in place that makes use of the language-aware fields, in addition to some extra testing, and will review at a later date...

[juntezhang]

Good catch! Adding more tests is definitely a good idea. Are you currently only working on this project?

…

On Wed, Apr 19, 2023, 11:35 AM Mike Bryant ***@***.***> wrote: It turns out that this issue was a bit more straightforward than anticipated, since we already have fields set up with the elision filter for various languages. We just weren't using them in this search 🙍‍♂️. This probably wouldn't have been noticed, except we added a "mimimum must match" parameter more recently, which in combination the language analysis setting for "generic" text, caused this search to fail. I have put a fix in place that makes use of the language-aware fields, in addition to some extra testing, and will review at a later date... — Reply to this email directly, view it on GitHub <#68 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAKDT2QD2734B2MNKQTTZNTXB6WVZANCNFSM6AAAAAAW2ISBVY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[mikesname]

Yes, way past time we had a proper comprehensive set of search tests, but no-one's had the time yet. I am not alone on this project thankfully (:wave: @mdermentzi @herminiogg ), though I'm responsible for most of the technical debt in this particular area!