Go with regexes. I wrote a small program which "normalizes" Delphi compiler messages. I use it to compare the messages of a previous build to the current build and fail if any warning/error was introduced. It has one regex to separate file, message id and message text and another one that parses the message text in a second step.
Will try, when I'm back at work in about two weeks. In theory it should, because the regexes contains no strings like "error" or "warning". I just grep for delimiters, text and numbers in general. They would fail right now if a different localization would use e.g. square brackets instead of round for the line numbers.
I am with David Schwartz here - use a regex to properly parse for the things that are the same in the error messages (the error code)
ReplyDeleteBut just for completeness sake I think removing the translation resources from the dccxx.exe and dccxx.dll should work.
Just set the corresponding registry entries in
ReplyDeleteHKEY_CURRENT_USER\Software\Embarcadero\Locales or HKEY_CURRENT_USER\Software\Borland\Locales.
Go with regexes. I wrote a small program which "normalizes" Delphi compiler messages. I use it to compare the messages of a previous build to the current build and fail if any warning/error was introduced. It has one regex to separate file, message id and message text and another one that parses the message text in a second step.
ReplyDeleteLübbe Onken and does it work regardless of the language?
ReplyDeleteWill try, when I'm back at work in about two weeks. In theory it should, because the regexes contains no strings like "error" or "warning". I just grep for delimiters, text and numbers in general.
ReplyDeleteThey would fail right now if a different localization would use e.g. square brackets instead of round for the line numbers.