[XML-SIG] SAX exceptions are odd

[Martin v. Loewis]

>   MvL> If you did provide a file name, and it got lost somewhere -
>   MvL> then that is a bug.
> 
> (It looks like you may have missed my second message on this subject.)
> I did pass a filename that was lost.

Actually, part of Germany was cut-off part of the US for the last day,
so some messages got sent later.

> I don't understand what you mean by reproducability.  The ability to
> reproduce an error message has nothing to do with whether it is terse
> or verbose.

If people write test suites, then they expact a certain output to
determine that a test failed. Any changes to the format of the output
will break the test case. The more verbose the text is, the more
likely are people to change it from release to release - not
considering that they may break things by changing some error message
formats.

Likewise, I expect that programs will parse the output of Python
programs, and expect a certain formatting. Such programs won't work if
people change strings.

> I liked the suggested error message that Lars proposed a *lot* better.

Reviewing the formats that Emacs' compilation-error-regexp-alist
supports, I found that <file>:<line>:<column>:error message is quite
common (and GNU standard), and that it would also recognize an
additional 'program name', so it correctly parses

xml.sax._exceptions.SAXParseException: a.c:10:16:not well-formed

(it does not parse the current text, as the error message precedes the
line information)

Regards,
Martin