another update to ovdb_server.c

[Heath Kehoe]

>On 23 Nov 2000, Heath Kehoe wrote:
>
>> http://www.avalon.net/~hakehoe/inn/ovdb2/ovdb_server.c
>
>Kewl.  Thanks.  Looks good under FreeBSD (execpt for the LOOPBACK
>address)

What's the problem with the loopback address again?  Is the bind()
failing?  What errno?


>
>MAP_ANONYMOUS is not known to FreeBSD, so I had to change it to
>MAP_ANON which I'm guessing is the same thing.  One could perhaps
>do, for portability, something like BerkeleyDB does, where it
>defines those MAP_ flags that are defined that it wants.
>

More #ifdef/#define type stuff will be added.


>It's been running for some hours with light readership and none
>of the original ovdb_swerver processes have had problems.
>
>
>All is not so well with Slowaris -- neither MAP_ANONYMOUS nor
>MAP_ANON are mentioned in the man page.

Are they defined by the headers?


>
>Also, it appears to exit quietly in the first round of the `for'
>loop around line 750 or so, the line that sets children[i].pid = 1
>appears to be at fault, so I haven't yet actually gotten it running
>yet with Solaris 2.7 (just compiled it now), so I can't say if
>there are any problems there.
>

If solaris can't do MAP_ANON, that would cause this.
I found a way to get around a lack of MAP_ANON[YMOUS]; and
I'll get an update out (along with the db-3.2.3 fixes)
after I get caught up again from being mostly AFC over the
last week. (AFC == Away From Computers)

-heath