[1.0RC23] fopen - bug or intended behaviour?

[Sslaxx]

So, in my code I have a line like so:

Code Select Expand

a = fopen ("Test.ini", O_READWRITE);

If Test.ini doesn't exist fopens returns 0 and attempting file operations causes a segfault. If using O_READ or O_ZREAD that would be understandable; the file doesn't exist. O_WRITE works (for writing the file out) as it creates the file if it does not already exist. O_READWRITE, as just described, behaves in the same way as O_(Z)READ.

I'd have thought, given what it is, that O_READWRITE would have ensured the file would be created if it does not already exist. Is the fact it does not do so intended behaviour, or a bug?

[josebita]

Not sure what the expected C equivalent behaviour is but as a general programming tip, it's always a good idea to ensure the file exists with file_exists() before trying to read from it.

[Sslaxx]

If you go by this page, then you're saying O_READWRITE is functionally the same as "r+" - "Open a file for update both reading and writing. The file must exist." Makes sense.

[josebita]

I just checked the BennuGD source code for mod_file and that's right, it's doing a "r+b" for O_READWRITE. (the "b" there is just a compatibility flag for old MS OSes)

[SplinterGU]

no, no is a bug.

for rw operation the file must be exists, if fopen return 0, it mean that the fopen operation fail, you must check this return code before continue.