starting compile server - advice needed

[Ziemowit Laski]

On Monday, Jul 14, 2003, at 15:39 US/Pacific, Matt Austern wrote:

> On Friday, July 11, 2003, at 04:26  PM, Per Bothner wrote:
>
>>> Should we autostart the server?
>>
>> That is the question I asked.  (I mainly asked *how*, but I also 
>> asked "should we".)
>
> I'd vote no.  It's easy enough for people who want to use this feature 
> to modify their makefiles, and a project maintainer understands 
> lifetime issues better than compiler heuristics can.

I'd vote yes. :-(  (I'm assuming that "autostarting" refers to starting 
up the server if it's not running already.)  I think making the compile 
server as transparent to the user as possible would be a win.  Also, 
any server parameters (whether to start the server, how long it should 
stay up, etc.) should be controlled by a few (no more than 2 or 3) 
environment variables.

If we force users to tweak their makefiles to take advantage of the 
compile server, I suspect that might lead to confusion that will easily 
dwarf the current PCH debacle. :-(

As always, my 2 zlotys...

--Zem

>
> I'd also vote for allowing users to start up multiple, named servers.  
> One major worry, if we don't do that, is that we might have several 
> projects trying to use servers at the same time.  Unless we're trying 
> to make gcc reentrant (yikes!), that'll mean locking and contention, 
> and it could end up hurting performance more than it helps.
>
> 			--Matt
>
>
--------------------------------------------------------------
Ziemowit Laski                 1 Infinite Loop, MS 301-2K
Mac OS X Compiler Group        Cupertino, CA USA  95014-2083
Apple Computer, Inc.           +1.408.974.6229  Fax .5477