66134 - nsDerivedSafe doesn't work on gcc 2.96 and gcc trunk

Status: RESOLVED FIXED

(Core :: XPCOM, defect, P1)

Description

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

nsDerivedSafe doesn't do what it's supposed to on gcc 2.96.
HAVE_CPP_ACCESS_CHANGING_USING is defined, so it uses the code:

private:
  using T::AddRef;
  using T::Release;

Probably the test for HAVE_CPP_ACCESS_CHANGING_USING should be changed (since
the code in nsCOMPtr.h is the only code that uses the result of that test).

This has given me a good bit of grief twice, and at some point I'll probably
turn the tree red because of it...

Comment 1

[robin shaw]

your probably already aware of this but, gcc 2.96 is broken, go to
http://gcc.gnu.org/gcc-2.96.html , for more info. 

Comment 2

[Scott Collins]

let's see if this works with gcc 3.0 (after getting the app to compile there)
and then decide if we need to fix this for 2.96 or not, given the non-release
nature of 2.96.  Targeting mozilla0.9.

Comment 3

[Scott Collins]

the new rule is that 2.96 is a known bad compiler which we don't support.
Marking this WONTFIX.  Sorry.  Please downgrade your compiler to 2.95.2 or see
if you can help make us compile under gcc 3.0.  Apologies for an unsatisfactory
answer.

Comment 4

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

This bug is also present on the gcc trunk as of 2-3 weeks ago.

Comment 5

[Scott Collins]

ugh, re-opening.  If it's a problem in the latest, then it really is a problem
that needs to be addressed.

Comment 6

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=3765

Comment 7

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Attachment: patch

Comment 8

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Taking bug.

Comment 9

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Comment on attachment 117411 [details] [diff] [review]
patch

Requesting reviews from seawood and bryner (configure.in patch), but scc is
also welcome to review.

Comment 10

[Brian Ryner (not reading)]

Comment on attachment 117411 [details] [diff] [review]
patch

sr=bryner

Comment 11

[hacker formerly known as [e-mail]]

Comment on attachment 117411 [details] [diff] [review]
patch

r=cls

Comment 12

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Fix checked in to trunk, 2003-03-17 19:45 PST.

Comment 13

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Oh, and the btek tinderbox log confirms that the test does still pass on egcs,
as it should:

checking whether the compiler can resolve const ambiguities for templates...
(cached) yes
checking whether the C++ "using" keyword can change access... yes
checking whether the C++ "using" keyword resolves ambiguity... (cached) no