On Tue, Feb 17, 2009 at 6:16 PM, Giovanni Bajo <[email protected]> wrote: > Any specific reason why you skipped it? The idea is that the additional > argument fail_on_invalid shouldn't be necessary, nor playing with the > "Not a valid URL" message. It is probably sufficient to check the return > value.
Well, the "Not a valid URL" part is for backward compatibility, as I remember it. The fail_on_invalid affects the behavior of get_svninfo -- whether to call error() or not. Most uses of get_svninfo want to error() out on an invalid URL, but check_url does not. Rather than adjust all calls to that function to conditionalize the result and call error() themselves, I added the extra parameter to control the function's behavior on failure. Obviously, the better solution would be to use exceptions instead of this error() business, but that's a big change for another day. I'm loathe to roll the first change back, only to have to re-debug something I debugged 9 months ago and re-apply this or something similar. As for the second change, this is the least intrusive fix in an already enormous patch. What do you think? Dustin -- Storage Software Engineer http://www.zmanda.com _______________________________________________ Svnmerge mailing list [email protected] http://www.orcaware.com/mailman/listinfo/svnmerge
