Re: [Patch] nsock/tests/tests_main.c

[Gisle Vanem <gvanem () broadpark no>]

"Henri Doreau" <henri.doreau () gmail com> wrote:

> I've committed the socket_strerror() part with a little change
> (r31677). Can you elaborate on why ANSI codes should be avoided? I
> wouldn't put them everywhere but they're something quite common for
> such tests.

On machines where you have an ANSI-driver in the console, yes.
The Windows console doesn't have that. Running this here now
produces these codes verbatim on the console (not colour changes):

g:\MingW32\src\inet\nmap\nsock\tests> tests_main.exe
nsock pool user data                            [←[1m←[32mOK←[0m]
test timer operations                           [←[1m←[32mOK←[0m]
set standard log levels                         [←[1m←[32mOK←[0m]
check error log levels                          [←[1m←[32mOK←[0m]
simple tcp connection                           Socket troubles: No error
Assertion failed: sd >= 0, file engine_select.c, line 353

The assert() is also a non-portable issue with WinPcap; it doesn't
have a selectable pcap structure or something.

Back to the colours on Windows. tests_main.c could have some
macros like:

#ifdef WIN32
 #define BOLDGREEN (FOREGROUND_GREEN | FOREGROUND_INTENSITY)

 /* default WinCon state */
 static CONSOLE_SCREEN_BUFFER_INFO csbi;
 static void init_console (void)
 {
  ...
 }
 #define PRINT_OK() \
    SetConsoleTextAttribute (stdout_hnd, (csbi.wAttributes & 0xf0) | BOLDGREEN); \
    printf ("OK\n"); \
    SetConsoleTextAttribute (stdout_hnd, csbi.wAttributes)
#else
 #define PRINT_OK() printf(TEST_OK "\n");
#endif

And use PRINT_OK() instead of printf(TEST_OK "\n").
Similar for TEST_FAILED. I could make a patch if there's an interest.

--gv

_______________________________________________
Sent through the dev mailing list
http://nmap.org/mailman/listinfo/dev
Archived at http://seclists.org/nmap-dev/