Re: flexfax: problems with protocol-rtn.patch

[Dmitry Bely <dbely@mail.ru>]

Hal Snyder <hal@vailsys.com> writes:

> This is mainly for Dmitry, but I'm posting it to the list in case
> anyone else has run into the same thing.
> 
> I've run into a couple problems compiling hylafax from CVS with
> protocol-rtn.patch, and wonder if there aren't a couple problems with
> the patch as it appears on the hylfax.org website. This is on
> FreeBSD-4.0, BTW.

The patch from hylafax.org was for unpatched beta 2 (not CVS
version). There is more recent patch against CVS (BTW, published two or
three weeks ago in hylafax-devel mailng list). Of course, it would be
better to apply the changes directly to CVS, but ...

> Line numbers apply to the CVS file after applying protocol-rtn.patch.
> 
> 1. In faxd/FaxModem.c++, line 649 - I think "runs" needs to be a
> uint32* rather than a uint16* - and the initialization around line 673
> also needs to be "runs = new uint32[2*width];" rather than using
> uint16. This is to keep compilation happy with the call to setRuns at
> line 676 - the latter method needs long int * arguments, not short
> int *.

Yes, if you use libtiff 3.5 and no, if you use 3.4. That's why I claim that 
this part of Hylafax/libtiff interface should be redesigned. In the
meantime I'll add there some conditional compilation directives.

> 2. In faxd/FaxModem.c++, line 735 and faxd/Class20.c++, line 179,
> there are casts (u_char*)result and (char*)s, respectively. I think
> these need to be explicit invocations of the (const char *) method for
> the object in question - certainly that was needed to get compilation
> on my platform.

Already done two weeks ago.

> Does this make sense?

Yes, of course :-)

> BTW, not relevant to Dmitry's patch -
> 
> 1. In util/Dispatcher.c++, around line 39, I have to #undef SA_NOCLDSTOP
> to get a compile on FreeBSD.
> 
> 2. In util/TypeRules.c++, there are capitalized instances of TRUE and
> FALSE on lines 107 and 109 that need to be changed to lowercase.

Hmm, are you absolutely sure that you are using current CVS version? There
is no capitalized TRUE at all in my copy of util/TypeRules.c++

BTW, don't you think we should move to hylafax-devel mailing list?

Hope to hear from you soon,
Dmitry