Re: flexfax: hylafax beta and Linux 2.2.x

[Robert Colquhoun <rjc@trump.net.au>]

At 11:03 PM 5/15/99 -0400, Dave Wreski wrote:
>> Can you cut and paste the warnings and errors.  fds_bits are used by the
>> FD_SET, FD_CLR macros i think, maybe something is going wrong there.
>
>These are the most relevant messages: (using the 2.2.9 kernel source)

oh....bleeding edge ;-)   What version of glibc?

>
>make[3]: Entering directory `/usr/src/hylafax-v4.0pl2-rjc6/util'
>/usr/bin/g++      -I.././zlib                     -D__ANSI_CPP__ -I. -I.. 
>-I../.
>/regex -I.././util -I.././util -I/usr/local/include -g -O  -c Dispatcher.c++
>In file included from Sys.h:31,
>                 from Dispatcher.c++:27:
>/usr/include/stdlib.h:566: warning: declaration of `mktemp(char *)' throws 
>diffe
>rent exceptions
>../port.h:34: warning: previous declaration here
>/usr/include/stdlib.h:573: warning: declaration of `mkstemp(char *)' throws 
>different exceptions

Could you look up the prototypes of these functions, either from
/usr/lib/stdlib.h or the man pages.  It looks like somehow hylafax has
decided to generate its own mktemp() and mkstemp() functions which are
conflicting with the standard ones provided by linux.

(This is the second mkstemp() problem i have seen today.... i am seriously
thinking of replacing all the mkstemp() calls with the posix tempnam()
function)


>../port.h:24: warning: previous declaration here
>In file included from Sys.h:35,
>                 from Dispatcher.c++:27:
>/usr/include/unistd.h:791: warning: declaration of `getpass(const char *)' 
>throw
>s different exceptions

Same as above i think, could you cut and paste the linux prototype.

>../port.h:25: warning: previous declaration here
>Dispatcher.c++: In method `fxBool FdMask::anySet() const':
>Dispatcher.c++:67: `fds_bits' undeclared (first use this function)
>Dispatcher.c++:67: (Each undeclared identifier is reported only once
>Dispatcher.c++:67: for each function it appears in.)
>Dispatcher.c++: In method `int FdMask::numSet() const':
>Dispatcher.c++:78: confused by earlier errors, bailing out
>make[3]: *** [Dispatcher.o] Error 1
>

This is more serious, it seems that hylafax has assumed some stuff about
the internals of the FD_* macros....ie how the operating system contructs
fd_sets.   This might need rewriting.

>> Attached below is a copy of Darrens email from a month or so ago:
>
>Ok, great, thanks.  Any idea how/if I can make this go thru a firewall?  BTW,
>has this CVS version been tested on RH 6.0 with 2.2.x?
>

Depends on your firewall....should not normally be a problem if you are
only running a cvs client(ie downloading the hylafax source code).  A cvs
server is another matter and probaly would require firewall configuration.

I am not sure if anyone has yet has tried RH 6.0, it is not available in
Australia until next week.

- Robert