monit-general
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: README v1.5 - darwin build note


From: Rick Robino
Subject: Re: README v1.5 - darwin build note
Date: Wed, 4 Jun 2003 03:59:55 -0700

Rory,

On Tuesday, June 3, 2003, at 11:40  PM, <address@hidden> wrote:



[...]
process/sysdep_DARWIN.c:96: warning: `calcru' defined but not used

Upon running the finished product the runtime linker (?) does give a
warning:


Thanks, I'll incorporate the patch. I've also downloaded the Darwin
version of top to give me some pointers on getting the /proc stuff
integrated (it's like no other...)

Don't bother - it is invalid, but probably not because of recent commits. Seems more likely that Including and Linking the standard system paths was not the issue after all, but applying the patch doesn't hurt (still works) and does result in a clean compile. That fact is a bit disturbing because I found that there were other solvable problems lurking in the Darwin port or my autoconf when I made this work without the configure.ac tweaks:

For starters, autoconf doesn't #define PACKAGE_VERSION in config.h (or anywhere else) for me, and monitor.h needs it so that VERSION gets a value. No one else has mentioned this so it must be me; I can't imagine how it couldn't have gotten noticed if it isn't. Anyway, I just added this under AC_INIT in configure.ac:
        AC_DEFINE([PACKAGE_VERSION], "3.3",
                [This release version.])

After that, I had show-stoppers on base64.o - needed "net.h" included; same thing with http/http_utils (needed net.h); immediately after common.o needed "alert.h" included by monit_process.h. Finally I got the same warnings at the end as last time about multiple definitions of _signal and that was it.

Again, I got a working monit binary that seemed to perform fine but did elicit the same ptr deallocation error as before. Interestingly this warning occurs for every invocation of monit besides "monit -h"...

So configure.ac doesn't need to be patched unless that PACKAGE_VERSION #define really is a bug. The rest I dunno... but after seeing what you stole from libtop already I do feel for you ;-) Are you looking at the OpenDarwin top sources now or just at FreeBSD? I wish I could help but like you I find this systems plumbing quite unique.

BTW, I ran MallocDebug (a neato apple tool) for Jan but nothing jumped out at me. Ditto for gdb with all the debug environmental vars the apple linker suggested. Got the warning at startup but then the app just stepped to completion without any noise. <shrug>... I'm sure a real NeXT wizard will drop by here eventually.

I've attached an annotated (****) log of my build from autogen through each fix and resumption of the make, plus the runtime warning to give you guys some visibility as it sounds like none of the core developers have a mac to play with (none at Savannah?). And also a cvs-diff showing the actual changes I made. I don't run a Mac server in production but judging from the night I've left it up I'd say it works fine on this platform. Knock on wood.

Cheers,

--

Attachment: rrobino-darwin-cvs.diff.gz
Description: GNU Zip compressed data

Attachment: rrobino-darwin-build.out.gz
Description: GNU Zip compressed data




reply via email to

[Prev in Thread] Current Thread [Next in Thread]