Open Krysztophe opened 1 year ago
darold
commented
1 year ago Hi Christophe,
About point 1, this is not an easy work, there are 14 language to support and I'm not sure that the Perl code will support Korean or Japanese words in the regexp.
The problem with point 2 is that this is common to have non-English language at start of the first log before lc_messages is changed.
Krysztophe
commented
1 year ago @gleu has launched a discussion about giving up the translation of server messages : https://www.postgresql.org/message-id/flat/CAECtzeV6dyu4jTOrorFW%3DB%3DEicyejWO7_Seew3Ch0%3D0wO%2BM-RQ%40mail.gmail.com
1) Making pgbadger aware of other languages would be tedious, but not impossible. We could support only the easiest languages at the beginning, and it would be easier to support the others.
2) Well, I'll try if I find some time for this.
''lc_messages = C'' is not set by default by some customers, so I have already lost time generating empty reports before checking it.
A question: can we imagine having pgbadger understand other languages? Some day I feel motivated, but it is probably a long and buggy task to find and substitute all keywords hardcoded in the regex.
Easier: add a non-blocking warning if some non-English language is detected. Perhaps a regex with the translations of the most common keywords from the translations, tested only on the first 1000 lines?