-
Bug
-
Resolution: Fixed
-
Minor
-
None
Change-set's Identity regex once broke against a Swedish character, for us here at Ericsson. By "broke" I mean that for some reason, our Jenkins/Java stack was running latin-1 character set/encoding, instead of utf8. So our git handled that Swedish character as utf8 (2-bytes) but Jenkins/Java read it as latin-1 (1-byte) -through that "broken" change-set Id regex. What’s too bad is that the character in that Swedish user name takes 2 bytes in utf8: \303 \205 (octal). But \205 doesn’t exist in latin-1, so the regex dot didn’t want it and the match failed.
Needed then is a robustness fix so that regex (GitChangeSet IDENTITY's) does not reject such a possibly-desired match. We say so even though there is no guarantee that such a fixed regex would keep matching our mistreated Swedish name in the future. All we know is that we once needed that fix to survive an awkward (yet realistic) Jenkins/Java stack setup (in terms of default locale and/or character-set encoding).