Hallo all,
Sorry for only replying now. We, South Africans, had a public holiday yesterday. Today is my first day back at the office.
Marcus:
Thank you for the patch in 7013. I got 7015 this morning and it did indeed make a big difference. The Magma Monticello load is still a bit slower than in 3.8, but it is a lot more acceptable now.
Chris:
Don't worry, I also have no idea how Magma works in 3.9 yet either (I don't even know how 3.9 works :-)). But it is always nice to try something new and there are already a couple of things that I like in 3.9 (or would like to explore).
After loading Magma, I couldn't get the unit tests to run completely. The test procedure gets stuck in the middle of 82 tests and when I alt-. , I see that it has something to do with LargeInteger>>+. It looks as if primitive 21 didn't work correctly. I am still very new to the primitive concept, but I will continue to dig around (if and when time permits) and see if I can figure out what actually is going on. (BTW: The call stack originates from MaHashIndexRecord>>byteArrayIndexesFrom:to:do:.)
Then there is also the deprecated method, but I am not sure how that affects everything yet.
Once again, thanks for everyone's help and comments.
Kind regards,
Mart-Mari
-----Original Message----- From: Petr [mailto:petr@madnetwork.org] Sent: 20 Maart 2006 03:29 To: magma@lists.squeakfoundation.org Subject: Re: For interest: Magma in 3.9a-7008
Hi Marcus, I did have a full Monticello installed. Petr
Marcus Denker wrote:
Hi,
Did you have Monticello installed in the 3.8 image? If not, then it
uses
MCInstaller for loading the package, which is much faster then
Monticello
itself (which is installed in 3.9a).
Other then that: Yes, it's slow... we are looking into it a bit as time permits. Friday we found a huge bottleneck of Monticello to be calling
#category
on Classes a lot, which does a linear search... 7013 has a patch that makes #category ~ 10 times faster on everage (for classes "at the end" which newly loaded classes are, much much more). This turns out to double MC loading speed... amazing.
I'm sure there are a lot of cases like this in Squeak... we only need to find and fix them ;-)
Marcus
On 17.03.2006, at 13:30, Mart-Mari Breedt wrote:
Hi,
Sorry about the re-asking. I only joined the new list today.
I had the same symptons... During most of the load there was no load
on
the CPU as well.
We can continue discussing this on squeak-dev. Perhaps someone on
that
side has an idea of what might be wrong.
Kind regards,
Mart-Mari
-----Original Message----- From: Petr [mailto:petr@madnetwork.org] Sent: 17 Maart 2006 02:20 nm To: magma@lists.squeakfoundation.org Subject: Re: For interest: Magma in 3.9a-7008
Hi, I asked this question here in my e-mail few days ago. I am mostly
using
the 3.9a images and when I tried to load Magma into 3.8, it was lightning fast. During most of the loading time, there is was no load
on
CPU. It happens with all packages in 3.9a loaded from Monticello. I guess we should discuss this issue in the squeak-dev.
Petr
Mart-Mari Breedt wrote:
Hallo all,
I am not sure whether this mail belongs on the Magma mailing list or
the
Squeak mailing list. Since my problem only seems to occur when installing Magma, I thought it better to post it here first.
This morning I was playing a bit with the new squeak image,
3.9a-7008.
I
found that installing Magma1.0-cmm.4.mcz (from a local copy) takes
about
twenty-five (!) minutes. The installation otherwise progresses fine, except for a warning towards the end of the installation: "CompiledMethod>>who has been deprecated. use #methodClass and
#selector
directly", but you can still proceed in this case. The same
installation
in 3.8-6665 doesn't even take a minute.
Maybe someone can venture a guess as to why the Magma installation
takes
so much longer in the new image?
Kind regards,
Mart-Mari
Magma mailing list Magma@lists.squeakfoundation.org http://lists.squeakfoundation.org/mailman/listinfo/magma
Magma mailing list Magma@lists.squeakfoundation.org http://lists.squeakfoundation.org/mailman/listinfo/magma
Magma mailing list Magma@lists.squeakfoundation.org http://lists.squeakfoundation.org/mailman/listinfo/magma
_______________________________________________ Magma mailing list Magma@lists.squeakfoundation.org http://lists.squeakfoundation.org/mailman/listinfo/magma
magma@lists.squeakfoundation.org