stat() for '\..' should fail

[KO Myung-Hun]

Hi/2.

stat() for '\..' should fail. A parent of root is invalid.

[KO Myung-Hun]

patch for '\..'

[bird]

No, the parent of root is root, on FreeBSD, Linux, Windows and most importantly OS/2. Play around with stat /../../.. if you don't believe me.

[KO Myung-Hun]

I've confirmed. But VAC 3.08, OpenWatcom v1.8 and EMX return an error. kLIBC should considers the compatiblity with other OSes more importantly than other OS/2 compilers ?

[bird]

Yes, POSIX/UNIX compatibility is much more important than being compatible with IBM VisualAge for C++, [Open]Watcom C/C++ or Microsoft C. Again just try call stat("/../../..", &St); on FreeBSD.

As mentioned in the wiki and numerous bugs, kLibC is NOT aiming at being bug compatible with either of the compilers you mentioned, but rather making porting UNIX open source code easier. This means leaning towards API behavior described by OpenGroup / POSIX and actual UNIX implementations or clones, rather than some obscure and mostly discontinued compilers.

(Seems the Dos* APIs on OS/2 are unable to resolve "C:\..", despite the root directory having a '..' entry. I recall there being a HPFS or HPFS386 (or was it JFS?) APAR adding it to the root directory a very long time ago, too bad they didn't make it do what it is supposed to do.)

[KO Myung-Hun]

I already confirmed on MingW and Linux. stat() worked on those OSes as you said. But OS/2 compilers did differently. So I asked.

I certainly came to know the aiming point of kLIBC. And I absolutely agree with you. Thanks.