On 05/01/2012 12:33 PM, Alan Coopersmith wrote:
On 05/ 1/12 12:24 PM, Rich Burridge wrote:
On 05/ 1/12 11:13 AM, Danek Duvall wrote:
When you did your testing, did you see the file descriptor leak prior to
adding the fix, and see that it went away afterwards, or was your test
simply that aasavefont didn't break?
You answered your own question with the comment below. The file
descriptor will be automatically closed as the program exits. Parfait
doesn't appear to have the smarts to understand that and deal accordingly.
Since parfait is just analyzing the library and not the callers of it,
Parfait can't tell how the library will be called by applications, or if
it's going to be a problem that it loses track of file descriptors it
opened.
In this case, the Parfait error isn't to a library; it's to a small
standalone
application. See:
/net/stard.us.oracle.com/tank/ws/UL/7164355/components/aalib/aalib-1.4.0/src/aasavefont.c
This is even in the main() routine. I really would of thought that it could
have been smart enough to work it out.
But anyhoo, it looks like we have a meta-issue here. Should we be fixing all
"problems" that parfait flags, irrespective of whether they really are
issues?
(In a couple of the other code review requests), it is indeed flagging false
positives, and I've "solved" it by adding an appropriate Parfait-style
comment
just before the line reporting the Error.)
Or should we only be fixing the ones that are indeed really problems?
_______________________________________________
userland-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/userland-discuss
