Fwd: Re: [Help-gnutls] Problem building from git

[Brad Hards]

On Monday 13 July 2009 17:50:01 Nikos Mavrogiannopoulos wrote:
> Brad Hards wrote:
> > On Monday 13 July 2009 16:39:30 Nikos Mavrogiannopoulos wrote:
> >> This is quite strange. In my copy x509.h:780
> >> gnutls_x509_crq_get_extension_by_oid has argument 5 as unsigned int and
> >> not size_t*. Is it the same in your system?
> >
> > No, it isn't. I have:
> >   int gnutls_x509_crq_get_extension_by_oid (gnutls_x509_crq_t crq,
> >                                         const char *oid, int indx,
> >                                         void *buf, size_t * sizeof_buf,
> >                                         unsigned int *critical);
>
> Sorry didn't check correctly. Does the following patch solve the issue
> for you?
There was one missing & (see previous message), and a lot of other problems. 
However the problems appear to mostly be one of two types:
1. Cases where we're using %d to print a ssize_t or size_t. That requires %zd 
instead. I'm confident about my fixes for those.
2. Cases where we are passing an integer to a function that expects a pointer 
(sometimes with, sometimes without a macro) to gnutls_transport_set_ptr().
This I'm less confident about, because (again) I'm not sure I understand the 
intent. Perhaps the address-of operator needs to be in the macro? Perhaps the 
macro needs to be used for the tests and examples as well?

There is something confusing (or confused) about tests/mini-eagain.c. It looks 
a bit like a work-in-progress.

Patch attached (zipped to get under the 40K limit on the list). Please provide 
review - I'm not very confident, and I got 9 out 
of 39 test failures.

Brad


-------------------------------------------------------

gnutls-build-2009-07-13.patch.zip
Description: Zip archive