The auto purge appears to be files and dirs. If yaws tries too hard to recover then we could encounter a recursive failure (sorry for the FUD)
I look forward to reading the patch.
On Oct 1, 2007, at 2:09 PM, Claes Wikstrom wrote:
PS: I think the code must always continue.... the first error message should be an error message pertaining to this err.
Yes, not enough badness to bail out. I changed the
error_logger message for this particular case. It still deserves
an entry in the error_log, since if/when it happens it means that
the local system is configured to auto purge logfiles in a way
yaws doesn't like.
Getting yaws to play nice with the normal way of rotating
logs, is an other story. Should/could be done, where by normal I mean
that at HUP/reload, all logfiles are simply reopened. That way a rotator can
rename files according to an external scheme and then HUP/reload yaws.
This is not how it works today though.