httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "William A. Rowe, Jr." <>
Subject RE: Tip for 2.0 Win32 Service code.
Date Fri, 16 Jun 2000 18:57:46 GMT
> From: []
> Sent: Friday, June 16, 2000 2:44 AM
> In a message dated 00-06-16 02:17:11 EDT, William Rowe writes...
> >  I am about 1 hour away from creating the service with
> >  the command argument:
> >  
> >   "path\apache.exe" --ntservice
> That'll work. I thought you had some aversion to doing that
> because of the complicated Apache argument handling or
> something. It's always been a viable option.

I did, because it did complicate things.  But I believe this
flavor of the solution I've committed addresses all my early
reservations.  In fact I've done something similar with 2.0,
which I now expect to change.

I'm trying to keep 1.3.x as similar as can be, so anything I 
did with the old services was ment to stay with the original
flow.  This patch does.  My 2.0 patch didn't (I let the getopt
parser pull out the -k runservice tags, which I no longer like.)

I have only one remaining issue, and I'm not really sure it's
an issue, with the 1.3.13 path.  The send_signal_to_service()
call passes on a return value, which if true, causes the server
to continue to start.  I don't think that it's playing nice.
But I can see using that return value to cause the Win9x path
to continue, without the extra strcasecmp of "start".  No issue
has been reported, I'm just trying to foresee every circumstance.

Here are the ViewCVS links to see the entire progress of the Win32
service changes in a nutshell (of course Greg's already seen it...
but other's aren't quite as familiar with all it can do :)

Start in the 6600's to see the actual 'service' changes,
but note there are Netware changes intermingled.

Go straight to the bottom and change the format from colored diff
to show whatever format is most astheticly pleasing to you.

Again, the service-related Win9x work is documented as highly
*experimental*, so I don't think we are doing anyone a diservice
by releasing these changes.  If you follow the NT flow, very, very
little has changed, and nothing should baffle the NT SCM.


View raw message