From owner-freebsd-ports Thu Jul 1 7:55: 6 1999 Delivered-To: freebsd-ports@freebsd.org Received: from smtp11.bellglobal.com (smtp11.bellglobal.com [204.101.251.53]) by hub.freebsd.org (Postfix) with ESMTP id A324D1578A for ; Thu, 1 Jul 1999 07:55:02 -0700 (PDT) (envelope-from hoek@FreeBSD.org) Received: from localhost.nowhere (ppp18334.on.bellglobal.com [206.172.130.14]) by smtp11.bellglobal.com (8.8.5/8.8.5) with ESMTP id KAA07296; Thu, 1 Jul 1999 10:57:51 -0400 (EDT) Received: (from tim@localhost) by localhost.nowhere (8.9.3/8.9.1) id KAA22587; Thu, 1 Jul 1999 10:55:42 -0400 (EDT) (envelope-from tim) Date: Thu, 1 Jul 1999 10:55:42 -0400 From: Tim Vanderhoek To: Neil Blakey-Milner Cc: Steve Price , Mikhail Teterin , ports@FreeBSD.org Subject: Re: tcl81 port takes forever to START building Message-ID: <19990701105542.B22489@mad> References: <199906291908.PAA74662@misha.cisco.com> <19990629213603.A49201@mad> <19990630001241.B70845@mad> <19990701152142.A82544@rucus.ru.ac.za> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.95i In-Reply-To: <19990701152142.A82544@rucus.ru.ac.za>; from Neil Blakey-Milner on Thu, Jul 01, 1999 at 03:21:42PM +0200 Sender: owner-freebsd-ports@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Thu, Jul 01, 1999 at 03:21:42PM +0200, Neil Blakey-Milner wrote: > > The reason for the long waits is the multiple "make" calls inside each of > the _PORT_USE macro users, like fetch, extract, &c. This isn't quite correct, but those do compound the problem. The tcl81 port is a bad port to look at because it twiggles some bug that is best either fixed or worked around - namely adding USE_X_PREFIX to the environment shouldn't make any difference in speed, but it actually makes an order or magnitude difference. In the case of 20+ second waits for tcl81, something causes make to spend a _lot_ (ie. 20+ seconds) inside the .for loop that constructs the _MLINKS variable. I currently have a patch that essentially takes this loop away and works around the bug (make has extremely slow .for evaluation anyways, bug or no bug, so it's a good idea regardless). > I think the only way to really fix it is to cordon off all the make calls > to sed, awk, and other crunchers when you're not doing something that > actually needs that work to be done. The problem is that you can't know when an individual port Makefile (eg. ports/xxx/yyy/Makefile) is going to require the results of one of those make calls. > If nothing's been done on this in two weeks, I'll consider myself > volunteered. (National Arts Festival/Grahamstown Festival is on) I have an 875 line message that I didn't send to the -ports list discussing slowness (parts of the problem were already being addressed before [MI] started jumping up and down). I didn't send it to the -ports list because 1) it's rather long, and 2) it includes patches that I'm not officially submitting yet (they'll get copied to -ports when/if that happens). I am, however, looking for comments on it and will gladly cc a copy to you or anyone else who thinks they actually have time to try and make changes and run some tests backing those changes up. You [Neil] just got yourself a copy. :) -- This is my .signature which gets appended to the end of my messages. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ports" in the body of the message