wessels [Mon, 7 Oct 1996 21:04:46 +0000 (21:04 +0000)]
From: Markus Gyger <mgyger@itr.ch>
OK, I couldn't resist to make some changes in cachemgr.c:
- URLs like .../cgi-bin/cachemgr?proxy.w3.org:3128 should now work
- .../cgi-bin/cachemgr?host=proxy&port=3128&url=http://www.x.org/&operation=
refresh
works like a POST request (parameters may be omitted)
- removed "Current" item in operation selection;
previous selection is remembered
- added "Empty Form" entry in selection
From: Markus Gyger <mgyger@itr.ch>
I've written a script for systems with SVR4 startup mechanism,
maybe it's useful for others too. However, I've tested it on
HP-UX 10.10 only, and there are some '?' where I don't know
what is used on those systems.
From: Markus Gyger <mgyger@itr.ch>
I've written a script for systems with SVR4 startup mechanism,
maybe it's useful for others too. However, I've tested it on
HP-UX 10.10 only, and there are some '?' where I don't know
what is used on those systems.
-Added pinger to Config.
-Pass debug_options as SQUID_DEBUG in environment to pinger.
-Have pinger send zero-length messages every 30s to make sure Squid is
alive.
From: webadm@cc.fh-lippe.de (Cord Beermann, Webmaster der FH Lippe)
Here is a small fix, which corrects the Cachemgr.cgi-Output for
non-table-Browsers.
From: Yoichi Shinoda <shinoda@jaist.ac.jp>
I guess we should not do strcasecmp(strtok(auth_hdr, " \t"), "Basic"))
in ftpStart(), because strtok() returns a null pointer when no more
token is available.
- Fixed coredump in icpStateFree() when calling
checkFailureRatio().
- Added more stuff to net_db.c.
- Fixed memory leak in stat_ipcache_get().
- Changed __P back to _PARAMS.
- Added 'cache_stoplist' to specify things which should
never get saved to disk.
- Merged some parsing routines into parseWordlist().
- Fixed icpHandleIMSReply() bug when HTTP reply headers
received in multiple reads.
- Changed all ipcache_nbgethostbyname() handler functions
to be void.
- Fixed bug when storeLockObject() fails for UDP_HIT_OBJ
replies.
- Simplified storeGet().
- Removed meta_data.hash_links member.
From: Markus Gyger <mgyger@itr.ch>
==============================================================================
> The most important change to this version is the switch to ANSI-style
> coding.
Great! Here are some diffs to keep my picky ISO C compiler more
silent.
From: Neil Murray <neil@aone.com.au>
I was doing some testing and found a case I hadn't allowed for in my
cachemgr.c mods.
Where you have a "?<proxy host>" without the ":" the code will play
up. The change is below and in case it is easier for you the entire
patch is below that. Assuming a virgin cachemgr.c from alpha 18 or
beta 1.
This also gives you the behaviour if that no ":<port>" is part of the
QUERY_STRING then it defaults back to the compiled port number, which
is what I think most users would expect. I could have made it so that
a "?:" defaults back to the compiled proxy hostname but decided that
this way "?:" will wipe any compiled hostname defaults and you are
presented with a blank set of dialog boxes.
- Added setting cachemgr.cgi fields from query string
(Neil Murray).
- Split log type TCP_IFMODSINCE into TCP_IMS_HIT and
TCP_IMS_MISS
- Replaced proto_hit() and proto_miss() with proto_count()
which looks at log_type argument to calcultate hits
and misses.
- Split CacheInfo into HTTPCacheInfo and ICPCacheInfo.
- Placed call to proto_count() just after log_append()
so we only need it in one place.
- Made a lot of proto.c functions static and added prototypes.
I'm sending you our Squid changes that Mark Kennedy discussed with you,
for optionally logging the full set of MIME request and reply headers
on every transaction. There are two sets of context diffs, one set for
1.1.10 and the other for 1.1.alpha12.
We'd love it if you bought these changes back; we think the log
information can be of great value and we'd love not to have to keep
retrofitting these changes into every release.
We've been running with this code essentially unchanged for several
weeks and it seems to be pretty stable.
From: Markus Stumpf <maex@Space.NET>
There *is* a problem with restarting the server :-(((
There is a
*list = NULL;
missing in acl.c at the end of aclDestroyDenyInfoList().
(This function is at the end of acl.c).