On Nov 22, 2009, at 2:56 PM, Anders Broman wrote:
No non-ASCII characters, please.
That happens occasionally when copy-pasting from standards documents
I don't know how to avoid it...
Unfortunately, the only way I know of is checking after editing. In
at least some UN*Xes, it *might* be possible to use "tr" to make a
copy of a file with all octets in the range 0x80-0xFF removed, and
then diff the two versions of the file, to see where the bad octets
were, although as the tr man page refers to "characters", it might
actually process the file using the current locale's character
encoding, so that a multi-octet encoding of a character is treated as
a single unit.
I seem to remember having trouble using tr, so I just cooked up a
small program to do the stripping of octets.
Gerald, can SVN be set up to run source files (.c, .h, etc.) through
the API filter at checkin time, rejecting them if they get errors?