This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: libsupc++ nit picks

Neil Booth <neil@daikokuya.demon.co.uk> writes:

| Nathan Myers wrote:-
| 
| > Cases of "if (p) free(p)" do represent mere paranoia.  
| 
| We still do this in cpplib, because as recently as mid-way through
| 2000 there were systems with broken C libraries that GCC was expected
| to work on (I think Kaveh may have an idea which), and when we weren't
| doing this things broke.

Same here.  When I started the work on diagnostics, I distinctly
remember having broken the tree on some variants of SunOS 4.x because
I didn't check p before free()ing it and the system library free() got
it wrong. 

|  I'd advise against changing this, unless we
| really don't care about such systems any more.

Completely agreed.

-- Gaby
Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]