Re: Preliminary work on jdk build time improvements

Brad Wetmore wrote: This is a great optimization. A couple comments. Platform.gmk: Haven't quite figured out why if MAKELEVEL=0 you export $1:=$2, but only export $1 if MAKELEVEL != 0. Sure there's a good reason why, doc'ing it would be appreciated. The $2 is the thing you don't want e

Re: Preliminary work on jdk build time improvements

This is a great optimization. A couple comments. Platform.gmk: Haven't quite figured out why if MAKELEVEL=0 you export $1:=$2, but only export $1 if MAKELEVEL != 0. Sure there's a good reason why, doc'ing it would be appreciated. 48/49: (if $(strip $1)... according to the gnumake docs:

Re: Should GNU make 3.81 be required?

I only mentioned the sum because the one Kelly was suggesting was probably the same one I got from cmake. I wasn't suggesting that you should be testing for that particular version. I would hope you can just create a simple check to see if the drive letter logic is on, possibly using cygpath.

Re: Should GNU make 3.81 be required?

Brad wrote: > Kelly O'Hair wrote: > > > I have a 3.81 for cygwin at http://cr.openjdk.java.net/~ohair/make.exe > > This binary sums to the same value as the one on cmake. > > > Brad "Spent way too much time on this in July" Wetmore Ow - my head hurts. So we need Windows GNU make version 3.81

Re: Should GNU make 3.81 be required?

It sounds like handling drive letters would be a good sanity test for make. -- Jon Brad Wetmore wrote: Tim wrote: > Excellent! We may be in better shape than I thought, unless there > are other make.exe downloads that report 3.81 yet are broken for > our purposes - that would be confusing.

Re: Should GNU make 3.81 be required?

Tim wrote: > Excellent! We may be in better shape than I thought, unless there > are other make.exe downloads that report 3.81 yet are broken for > our purposes - that would be confusing. A quick reminder, the gnumake 3.81 as part of a normal cygwin install does *NOT* support the drive letter

Re: Should GNU make 3.81 be required?

I have managed to build a GNU make 3.81 on Solaris 8 (x86 & sparc), and on the old Linux systems we use for jdk5 and jdk6. I have a 3.81 make.exe for cygwin that works, so all I'm missing at this point is a gnumake.exe that will work with MKS builds. GNU make builds very easy on *nix systems (./c

Re: Should GNU make 3.81 be required?

David Holmes wrote: > On Solaris 10 /usr/sfw/bin/gmake (which IIRC comes from the Solaris > companion CD) is 3.80. Correct - anything in /usr/sfw or /opt/sfw is from a Software Companion CD. > /java/devtools/sparc/bin/gnumake is 3.78.1 > > I presume the devtools version will be updated to 3.81

Re: Should GNU make 3.81 be required?

Hi Kelly, Kelly O'Hair said the following on 09/16/09 03:24: So I'm trying to work out some jdk7 Makefile performance improvements and everything is working out fine on Linux and OpenSolaris... Then I discover that the GNU make features I'm using are seriously broken in GNU make 3.80 and I need

Re: Need reviewers - 6856630: Restructure jaxp/jaxws repository

Kelly, I'm downloading the JAXP bundle from the kenai test site. It was very slow initially (almost not moving). But it became normal after I killed the 1st attempt. I'm not sure if that's what you meant when you mentioned kenai download was not stable. Maybe it was my internet connection. Di

Re: Should GNU make 3.81 be required?

I wrote: >> I am currently using Cygwin GNU make v3.80 on Windows due to the >> later Cygwin builds not supporting MS-DOS drive letters or paths: >> >> http://hg.openjdk.java.net/jdk7/jdk7/raw-file/tip/README-builds.html#gmake Kelly O'Hair wrote: > I have a 3.81 for cygwin at http://cr.openjdk

Re: Should GNU make 3.81 be required?

Mark Reinhold wrote: Date: Tue, 15 Sep 2009 10:44:08 -0700 From: [email protected] For as long as I remember, we've used make 3.78.*. I assume the most problematic platform to upgrade is windows. However, if there are significant benefits build time benefits, I think upgrading the make ver

Re: Should GNU make 3.81 be required?

So do I continue on, and assume a new requirement to upgrade to GNU make 3.81 because of these changes (on all platforms), or do I go back to the drawing board and try and make the changes work with 3.80, or even 3.78.1 ??? If we can find a Windows/Cygwin GNU make v3.81 or better that works, I

Re: Should GNU make 3.81 be required?

Tim Bell wrote: Hi Kelly: So I'm trying to work out some jdk7 Makefile performance improvements and everything is working out fine on Linux and OpenSolaris... Then I discover that the GNU make features I'm using are seriously broken in GNU make 3.80 and I need 3.81 to get around it. :^( Tha

Re: Should GNU make 3.81 be required?

> Date: Tue, 15 Sep 2009 10:44:08 -0700 > From: [email protected] > For as long as I remember, we've used make 3.78.*. I assume the most > problematic platform to upgrade is windows. However, if there are significant > benefits build time benefits, I think upgrading the make version is justified

Re: Should GNU make 3.81 be required?

Hi Kelly: > So I'm trying to work out some jdk7 Makefile performance improvements > and everything is working out fine on Linux and OpenSolaris... > Then I discover that the GNU make features I'm using are seriously > broken in GNU make 3.80 and I need 3.81 to get around it. :^( That's painful.

Re: Should GNU make 3.81 be required?

Kelly O'Hair wrote: So I'm trying to work out some jdk7 Makefile performance improvements and everything is working out fine on Linux and OpenSolaris... Then I discover that the GNU make features I'm using are seriously broken in GNU make 3.80 and I need 3.81 to get around it. :^( Also it turns

Should GNU make 3.81 be required?

So I'm trying to work out some jdk7 Makefile performance improvements and everything is working out fine on Linux and OpenSolaris... Then I discover that the GNU make features I'm using are seriously broken in GNU make 3.80 and I need 3.81 to get around it. :^( Also it turns out that GNU make 3.