Re: squid 1.15b OS/2 - dnsserver won't start

From: Duane Wessels <wessels@dont-contact.us>
Date: Tue, 24 Nov 1998 18:04:02 -0700

"Doug Nazar" writes:

>On Tue, 17 Nov 1998 17:50:09 -0600 (CST), Timur Tabi wrote:
>
>>Interest?!?!!?!? I'll name my first born after you, even if it's
>>a girl! (she'll understand)
>
>Sorry, for taking so long to get back to you but I'm just now starting
>to recover from a cold or something. I've also got most of my stuff
>packed up to move this Sunday.
>
>>>I'm not sure if I want to merge the changes back into the mainstream
>>>since it's mainly working around bugs in emx.
>>
>>I think it would be nice if you did - that would improve the change of
>>version parity between Linux and OS/2.
>
>After spending some time with the squid & emx code, I'm starting to
>wonder if my time won't be better spent improving the emx library. Most
>notably, sockets aren't inherited across exec*() calls.
>
>One of the other main issues is that's you can't rename (or unlink)
>open files. Unfortuanatly I'm having a little trouble following the
>code paths for handling the file handle closing.
>
>If I call file_close() and flags.write_daemon is set, then the close
>will be delayed but I'm not sure what to call to force the close
>properly. This is causing a race condition on startup & shutdown for
>the cache logs (swap.state, swap.state.new, swap.state.clean,
>swap.state.last-clean). Anybody?

If flags.write_daemon is set then there is more data to write
to the file. You probably don't want to close it before
all has been written.

If you must, I guess you could do something ugly like
call fd_table[fd].write_handler(fd, fd_table[fd].write_data);

I'm not sure what will happen next though.

Duane W.
Received on Tue Nov 24 1998 - 17:46:37 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:43:19 MST