Farid Zaripov wrote:
-----Original Message-----
From: Martin Sebor [mailto:[EMAIL PROTECTED]
Sent: Wednesday, August 02, 2006 8:30 PM
To: [email protected]
Subject: Re: testsuite process helpers [PATCH]
* rw_process.h (rw_pid_t): The type long changed to intptr_t
(on Windows) and pid_t (on other platforms).
I would like to hide the actual type and avoid #including
system (including the C library) headers in test suite
headers whenever possible. In this case, will _RWSTD_SIZE_T
do for rw_pid_t?
rw_pid_t should be signed, and should be 64 bit on 64-bit platforms.
I see that _RWSTD_SSIZE_T can be used.
Right. Or _RWSTD_PTRDIFF_T. They are going to be the same.
Is this enough? Don't we want to suppress the output of rw_note() (and
others as well)?
If so, I think a more robust solution might be for driver.cpp to
provide and handle a
variable controlling whether to issue or suppress a
diagnostic (and which). Something like int rw_diag_mask each
of whose bits would disable the corresponding diagnostic.
I found that variable is already exist (_rw_diag_mask in opt_trace.h).
Oh yeah, I forgot about that one. It gets set in response to
the --severity=<...> command line option. I'll need to check
to make sure it does what we want.
We can declare it
in driver.h (with removing the leading underscore of course) but it is
still not exactly what
is needed. When any diagnostic are masked it still counted and is
present in totals.
Ah, right. So it sounds like we need another mask. One that
prevents the diagnostic not just from being issued but also
from being counted.
But
in the case of 0.process.cpp test we need temporary disable the all
diagnostic messages
with disabling the counting ot that messages.
Right.
Another point: if we want to disable some type of diagnostic using
rw_diag_mask it is
needed to move the enum diag_t from opt_diags.h to driver.h too.
Yes, I suppose you're right. Unless we do some magic with
rw_error() and other diagnostic functions:
rw_error(0, 0, 0); // disables counting errors
rw_error(0, 1, 0); // disables issuing
rw_error(1, 0, 0); // enables counting
rw_error(1, 1, 0); // enables counting and issuing
Each call returns the previous state of the diagnostic setting.
Or something like that.
I propose add two functions enable_diag() / disable_diag(), or just
enable_diag(bool enable)
to the driver (and use any unused bit in _rw_diag_mask). We can publish
the _rw_diag_mask
along with enum diag_t later when it's will be required.
That would be another option. Which one of the two approaches
do you like better?
Martin
