1.1.4 incompatible with glibc 2.1

1.1.4 incompatible with glibc 2.1

[Sergey Suleymanov]

timid.c: In function `timid_preinit':
timid.c:105: `in_addr_t' undeclared (first use in this function)
timid.c:105: (Each undeclared identifier is reported only once
timid.c:105: for each function it appears in.)


-- 
  Sergey Suleymanov

Re: 1.1.4 incompatible with glibc 2.1

[Stas Sergeev]

Hello.

Sergey Suleymanov wrote:
> timid.c: In function `timid_preinit':
> timid.c:105: `in_addr_t' undeclared (first use in this function)
Of course you can change "sizeof(in_addr_t)"
to something like "sizeof(*serv->h_addr)",
but it is unclear why it doesn't work for
you.
"man inet" refers to "in_addr_t" a lot, so
it must be perfectly OK to use IMO.

Re: 1.1.4 incompatible with glibc 2.1

[Sergey Suleymanov]

>>>>> Stas Sergeev writes:

 >> timid.c: In function `timid_preinit': timid.c:105: `in_addr_t'
 >> undeclared (first use in this function)

 Stas> Of course you can change "sizeof(in_addr_t)" to something like
 Stas> "sizeof(*serv->h_addr)", but it is unclear why it doesn't work
 Stas> for you.  "man inet" refers to "in_addr_t" a lot, so it must be
 Stas> perfectly OK to use IMO.

        Glibc-2.2 has

typedef uint32_t in_addr_t;

        in /usr/include/netinet/in.h. Glibc-2.1 hasn't. :(

-- 
  Sergey Suleymanov

Re: 1.1.4 incompatible with glibc 2.1

[Stas Sergeev]

Hello.

Sergey Suleymanov wrote:
> Glibc-2.2 has
> typedef uint32_t in_addr_t;
> in /usr/include/netinet/in.h. Glibc-2.1 hasn't. :(
Yes, I understand that this error means
a missing definition. What am I trying to
say is that it looks like a bug of a
particular glibc version, so that no
modifications to dosemu required at all,
just "upgrade your glibc" message.
We can always hardcode that size to 4,
but what if midid will use IPv6 one day? :)
Could you, btw, please search for that
definition in the other headers? I can't
beleive it can be simply missing, after
reading "man inet".
It would be the best to add an in_addr_t
check to configure and define it ourselves
if it is missing, but probably this is just
too much for such a small thing:)

Re: 1.1.4 incompatible with glibc 2.1

[Sergey Suleymanov]

>>>>> Stas Sergeev writes:

 Stas> Yes, I understand that this error means a missing
 Stas> definition. What am I trying to say is that it looks like a bug
 Stas> of a particular glibc version, so that no modifications to
 Stas> dosemu required at all, just "upgrade your glibc" message.  

        Wow! :) That's sort of using a steam-hammer to crack nuts.

 Stas> We can always hardcode that size to 4, but what if midid will
 Stas> use IPv6 one day? :) Could you, btw, please search for that
 Stas> definition in the other headers?

        Nothing.

 Stas> I can't beleive it can be simply missing, after reading "man
 Stas> inet".  

        I read it too. And saw just unsigned long int :) System is
        Debian 2.2 (Potato)

 Stas> It would be the best to add an in_addr_t check to configure and
 Stas> define it ourselves if it is missing, but probably this is just
 Stas> too much for such a small thing:)

        :) Well, 1.1.4 is a development release. I think when 1.2 is
        out glibc will be 5.x ;)

-- 
  Sergey Suleymanov

Re: 1.1.4 incompatible with glibc 2.1

[Stas Sergeev]

[-- Attachment #1: Type: text/plain, Size: 507 bytes --]

Hello.

Sergey Suleymanov wrote:
> Stas> dosemu required at all, just "upgrade your glibc" message.  
> Wow! :) That's sort of using a steam-hammer to crack nuts.
Yes:)

> Stas> I can't beleive it can be simply missing, after reading "man
> Stas> inet".
> I read it too. And saw just unsigned long int :) System is
That makes sense.
OK, as the in_addr_t is not really used
(only once for sizeof), then there should
be no problems at all avoiding it entirely.
The attached trivial patch should do the trick.

[-- Attachment #2: timid.diff --]
[-- Type: text/plain, Size: 498 bytes --]

--- src/arch/linux/dosext/sound/midid/timid.c	Sat Jan  4 15:40:30 2003
+++ src/arch/linux/dosext/sound/midid/timid.c	Sun Jan  5 13:04:50 2003
@@ -102,7 +102,7 @@
   
   ctrl_adr.sin_family = AF_INET;
   ctrl_adr.sin_port = htons(config.timid_port);
-  memcpy(&ctrl_adr.sin_addr.s_addr, serv->h_addr, sizeof(in_addr_t));
+  memcpy(&ctrl_adr.sin_addr.s_addr, serv->h_addr, sizeof(ctrl_adr.sin_addr.s_addr));
 
   data_adr.sin_family = AF_INET;
   data_adr.sin_addr.s_addr = ctrl_adr.sin_addr.s_addr;