[hylafax-devel] Re: missing parts of security-patch.sh

[Andy Sparrow <andy@geek4food.org>]

--------
Your message dated: Wed, 21 Jun 2000 20:22:19 +1000
>At 06:04 PM 6/18/00 -0700, Andy Sparrow wrote:

>ok, you hylafax param Fontmap and Fontdir should be set to the same as the 
>ghostscript search path:
>         Fontmap:. : /usr/local/share/ghostscript/6.01/lib 
>:/usr/local/share/ghostscript/fonts
>         Fontdir:. : /usr/local/share/ghostscript/6.01/lib 
>:/usr/local/share/ghostscript/fonts
>
>so that textfmt et al find the same fonts that ghostscript will use to 
>rasterize the documents.

Well, it seems ironic then that 'afm' seems to have been broken out as a 
package at least partly for the benefit of Hylafax - from '.../afm/pkg/DESC':

  Adobe Font Metrics. Describes the sizes of different postscript
  types. Used by FlexFAX among others.

And the MASTER_SITE definition in the Makefile is:

  MASTER_SITES=   ftp://sgi.com/sgi/fax/source/

>> >> >...are you sure the above isn't the 'size_t' type, that seems to be what
>> >> >alot of the other platforms are using.
>><SNIP>
>> >Interesting they changed the type and it looks like they have a defined
>> >socklen_t to hide that from the user(which is what hylafax tries to use)...
>> >What does the manpage for accept give as the function prototype ie is the
>> >last param of socklen_t type?
>>
>>Yes it is.
>
>Hylafax should then work by default, without any changes to 
>configure...what error message you get if you remove the freebsd entries?

It doesn't cause any error message. I didn't compile on 4.x-STABLE
originally.

I scanned this shell code quickly:

    #
    # Some vendors have changed the socket API so that
    # call-by-reference parameters that give buffer
    # lengths are not int; this must be dealt with to 
    # get the code to compile.  Note that we only emit
    # the define if the type is not an int; we do this
    # so Socket.h can include code that assumes the
    # default so folks don't have to rerun configure.
    #
    test "$CONFIG_SOCKARGLENTYPE" || {          # int vs unsigned long vs ...
        case $TARGET in
        # XXX fill in for busted systems
        *-univel-*)     CONFIG_SOCKARGLENTYPE=size_t;;
        *sysv4.2uw2*)   CONFIG_SOCKARGLENTYPE=size_t;;
        *-UnixWare*)    CONFIG_SOCKARGLENTYPE=size_t;;
        *-aix4*)        CONFIG_SOCKARGLENTYPE=size_t;;
        *)              CONFIG_SOCKARGLENTYPE=int;;
        esac
    } 
    
Noted the comments and figured that anything that was an 'int' should be 
fine because that's what it defaults to using (e.g.FreeBSD 3.x) and anything 
that wasn't (e.g. FreeBSD 4.x and 5.x) should have a declaration setting 
CONFIG_SOCKARGLENTYPE to the appropriate type (i.e. u_int32_t).

However, you're right: it DTRT without the patch

Cheers,

AS


____________________ HylaFAX(tm) Developers Mailing List ____________________
 To unsub: mail -s unsubscribe hylafax-devel-request@hylafax.org < /dev/null