r8616: candidate for 1.0.1: Improved warnings from svnlook and svnadmin

I'm vetoing this for the moment, pending a modification or a good explanation :-)

1. The wrapper error is not cleared, so it leaks memory. The (only existing) caller of the warning call-back function clears the error that it passed in, but the warning function in this patch generates a new wrapper error and does not clear that. Unless my understanding of nested errors is flawed, that's a leak.

2. I don't see the point of adding the words "The filesystem warning function was called" to whatever error occurs. Information about which functions are called is only relevant to Subversion developers. I assume the message was intended to emphasise that this is only a warning, not fatal. If so, it could be reduced to something like "Warning:" or "Filesystem warning:", but I would be happy with no wrapper message at all unless you think that would be too confusing.

I know that point (2) is rather trivial, and that point (1) is not much of a practical problem given that the affected programs are not intended to be long-running, but we're trying to apply higher than usual standards to these candidates for back-porting.

- Julian

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Feb 20 00:49:20 2004