Can someone with a lot of RAM test this PKGBUILD?

wuischke · 2008-01-30 10:50:52

If someone could test this PKGBUILD I would appreciate it a lot. Please be aware that it will use a lot of RAM (about 2GB) during the process, which made my computer unusable during the process. I have only one PC here (with 1GB), so I had to cancel the build in order to use it. Everything should build fine, but I prefer to test things before I put them in AUR.

Maybe someone could create a package for x86_64 as well, if not I'll build it during the night.

# Contributor: Johannes Krampf <wuischke@amule.org>
pkgname=gf
pkgver=2.9
pkgrel=1
pkgdesc="The Grammatical Framework (=GF) is a grammar formalism based on type theory."
arch=('i686' 'x86_64')
url="http://www.cs.chalmers.se/~aarne/GF/"
license=('GPL')
makedepends=('ghc')
options=('!makeflags')
source=(http://www.cs.chalmers.se/~aarne/GF/download/GF-$pkgver.tgz)
md5sums=('0cbcd3de7835fbb3716699d817b130ec')

build() {
  cd $startdir/src/GF-$pkgver/src
  ./configure --prefix=/usr --enable-java=no
  make || return 1
  make DESTDIR=$startdir/pkg install || return 1
}

Last edited by wuischke (2008-01-30 11:01:40)

byte · 2008-01-30 10:56:50

Just a small remark: pkgname is lowercase.

Gilneas · 2008-01-30 10:58:15

Why does compiling this take 2GB?

wuischke · 2008-01-30 11:02:05

byte: Thank you, I've changed this.

Gilneas: I think this is due to recursive functions, but I'm not sure.

Zer0 · 2008-01-30 18:18:12

Make a temporary 2 gig swap file.. maybe that will work

dd if=/dev/zero of=/tmp/extraswap bs=1M count=2048
mkswap /tmp/extraswap                       
swapon /tmp/extraswap

Edit: nm I just read your other post..

Last edited by Zer0 (2008-01-30 18:20:10)

lilsirecho · 2008-01-30 19:04:48

If I run this, do I need to lower case the GF? I have 3GB.

wuischke · 2008-01-30 20:06:47

I've adjusted the above PKGBUILD already, so no changes should be necessary.

lilsirecho · 2008-01-30 23:46:27

Performed the package build for about a half-hour. It reached 1.75MB of ram in use at 29 mins. Cpu was @ 100% at all times.

The package aborted @~34 mins of activity withe the following error messgae.......

gf: internal error: evacuate: strange closure type 58951
(GHC version 6.8.2 for i386_unknown_linux)
Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug
make[2]: *** [alltenses] Aborted
make[2]: Leaving directory `/root/Desktop/Untitled/src/GF-2.9/lib/resource'
make[1]: *** [new] Error 2
make[1]: Leaving directory `/root/Desktop/Untitled/src/GF-2.9/lib/resource'
make: *** [lib] Error 2
==> ERROR: Build Failed.
Aborting...
[root@n6re Untitled]#
-------------------------------------------------------------
I had to download ghc and install @ 55MB...

Better luck next time!!

wuischke · 2008-01-30 23:54:52

Thanks anyway.

I'm just puzzled about the RAM usage. For me it would use a lot of CPU until it hit my physical RAM limit, then it would use 2-3%, but the PC was unusable.

I'll see if it works over night.

Snowman · 2008-01-31 02:24:34

package here: http://dev.archlinux.org/~eric/gf-2.9-1 … pkg.tar.gz

Your PKGBUILD wasn't working. the fixed one is below:

pkgname=gf
pkgver=2.9
pkgrel=1
pkgdesc="The Grammatical Framework (=GF) is a grammar formalism based on type theory."
arch=('i686' 'x86_64')
url="http://www.cs.chalmers.se/~aarne/GF/"
license=('GPL')
makedepends=('ghc')
options=('!makeflags')
source=(http://www.cs.chalmers.se/~aarne/GF/download/GF-$pkgver.tgz)
md5sums=('0cbcd3de7835fbb3716699d817b130ec')

build() {
  cd $startdir/src/GF-$pkgver/src
  ./configure --prefix=/usr --enable-java=no
  make || return 1
  make bindir=$startdir/pkg/usr/bin datadir=$startdir/pkg/usr/share install-gf install-gfdoc install-lib 
}

wuischke · 2008-01-31 07:16:13

Thanks a lot! I appreciate it.

May I ask you how long it took you to build it and what machine you are using? I tried to compile over night, but it was killed after about 4.5h of runtime. (I issued a "date" after the build command to see how long it would take...)

Jesus, is swapping on Linux really that bad? It's only a notebook-hdd, but it's relatively fast nonetheless. I think I'll do some testing with recursive functions to see if that's really the reason or if it's gf's fault.

Snowman · 2008-02-01 00:58:04

It took 12 min. Machine spec: AMD Athlon(tm) 64 X2 Dual Core Processor 5600+, 4GB RAM, 2GB swap. Near the end, it was using 98% of my RAM with 62% of my swap. It probably aborts on your machine because it ran out of swap space. I'm afraid it won't be able to build on your machine even if you would increase your swap. In the case it builds, it will probably take a very long time because of the swapping: in 4.5h you weren't able to build it while I did it in 12 minutes.

wuischke · 2008-02-01 13:11:21

Thanks...I guess there goes my PS3 (well, at least for now) and comes a new PC with a lot of RAM instead...

My current PC is probably a bit slower than yours (Core 2 Duo Mobile 1.6Ghz), but I, too, think that it's probably a RAM issue and not so much a question of processing power.

Arch Linux

#1 2008-01-30 10:50:52

Can someone with a lot of RAM test this PKGBUILD?

#2 2008-01-30 10:56:50

Re: Can someone with a lot of RAM test this PKGBUILD?

#3 2008-01-30 10:58:15

Re: Can someone with a lot of RAM test this PKGBUILD?

#4 2008-01-30 11:02:05

Re: Can someone with a lot of RAM test this PKGBUILD?

#5 2008-01-30 18:18:12

Re: Can someone with a lot of RAM test this PKGBUILD?

#6 2008-01-30 19:04:48

Re: Can someone with a lot of RAM test this PKGBUILD?

#7 2008-01-30 20:06:47

Re: Can someone with a lot of RAM test this PKGBUILD?

#8 2008-01-30 23:46:27

Re: Can someone with a lot of RAM test this PKGBUILD?

#9 2008-01-30 23:54:52

Re: Can someone with a lot of RAM test this PKGBUILD?

#10 2008-01-31 02:24:34

Re: Can someone with a lot of RAM test this PKGBUILD?

#11 2008-01-31 07:16:13

Re: Can someone with a lot of RAM test this PKGBUILD?

#12 2008-02-01 00:58:04

Re: Can someone with a lot of RAM test this PKGBUILD?

#13 2008-02-01 13:11:21

Re: Can someone with a lot of RAM test this PKGBUILD?

Board footer