Hello,
Id this just me, or is the VM on Linux (2.0.36, different machines, different X-servers) quite unstable. With the downloaded binary, I could send the VM in a closed loop by choosing the nextPage button in the Morphbook (one of the play-with-me's). I recompiled, and now nextPage works, but, for example, selecting '1 + abc' and doing it in the Workspace does not produce an error, but causes the VM to freeze. If I kill it via the WM button close, it gets in a closed loop. Playing around for a bit with the morph interface, is also almost guaranteed to get the VM in a closed loop.
It would be nice if someone could confirm this behaviour, or maybe there is something wrong with my system(s).
Linux-2.0.36, fresh, hardly updated, RedHat 5.2, with KDE desktop. (I tried different Xservers and WM's, but no difference there).
Han Holl
On Tue, 22 Dec 1998, Han Holl wrote:
Hello,
Id this just me, or is the VM on Linux (2.0.36, different machines, different X-servers) quite unstable. With the downloaded binary, I could send the VM in a closed loop by choosing the nextPage button in the Morphbook (one of the play-with-me's). I recompiled, and now nextPage works, but, for example, selecting '1 + abc' and doing it in the Workspace does not produce an error, but causes the VM to freeze. If I kill it via the WM button close, it gets in a closed loop. Playing around for a bit with the morph interface, is also almost guaranteed to get the VM in a closed loop.
It would be nice if someone could confirm this behaviour, or maybe there is something wrong with my system(s).
Linux-2.0.36, fresh, hardly updated, RedHat 5.2, with KDE desktop. (I tried different Xservers and WM's, but no difference there).
Intel linux? I built and packaged squeak this weekend, and I see the same behavior on my mklinux for powerpc installation. I just assumed that it was a peculiarity of mklinux. The MacOS version works beautifully.
By the way, I built an rpm for squeak which can be obtained from my machine at
ftp://bacon.ne.mediaone.net/pub/redhat/RPMS/ppc/squeak-2.2-1.ppc.rpm ftp://bacon.ne.mediaone.net/pub/redhat/SRPMS/squeak-2.2-1.src.rpm
The source package should build and install on other architectures than ppc. I packaged up a number of squeak html pages for the documentation portion of the package. The image, changes and sources files are put in /usr/local/share/squeak. A script called "newsqueak" will create a new squeak directory in your home directory and copy the image, changes and sources there. This makes it easy to keep the originals intact for multiple users.
Fred Bacon
========================================================================== Aerodyne Research, Inc. Phone: (978) 663-9500 ext. 273 45 Manning Rd. FAX: (978) 663-4918 Billerica, MA 01821-3976 http://www.aerodyne.com ========================================================================== Me spill chucker woks grate. I needle gramma chicken.
Fred Bacon wrote: [ cut ]
Intel linux? I built and packaged squeak this weekend, and I see the same behavior on my mklinux for powerpc installation. I just assumed that it was a peculiarity of mklinux. The MacOS version works beautifully.
Yes, Intel, sorry for forgetting to mention that. It would seem that this is a generic Linux (Unix??) bug.
By the way, I built an rpm for squeak which can be obtained from my machine at
ftp://bacon.ne.mediaone.net/pub/redhat/RPMS/ppc/squeak-2.2-1.ppc.rpm ftp://bacon.ne.mediaone.net/pub/redhat/SRPMS/squeak-2.2-1.src.rpm
Thanks. I'll have a try.
Han Holl
Linux-2.0.36, fresh, hardly updated, RedHat 5.2, with KDE desktop. (I tried different Xservers and WM's, but no difference there).
What C compiler are they bundling with 5.2? I do not see the DoIt hang that you mention when I use a VM compiled for Squeak 2.2 using kernel 2.0.33, libc-5.4.46, and gcc 2.7.2.3. Somehow the adjectives bleeding edge and production do not live comfortably together in my mind.
Christopher Oliver wrote:
Linux-2.0.36, fresh, hardly updated, RedHat 5.2, with KDE desktop. (I tried different Xservers and WM's, but no difference there).
What C compiler are they bundling with 5.2? I do not see the DoIt hang that you mention when I use a VM compiled for Squeak 2.2 using kernel 2.0.33, libc-5.4.46, and gcc 2.7.2.3. Somehow the adjectives bleeding edge and production do not live comfortably together in my mind.
RedHat is quite conservative with the C compiler (they want to be able to compile the kernel <g>). gcc-2.7.2.3-14. (For C++ they use egcs). I revisited VM 2.2, and now i get a proper doesNotUnderstand when I do '1 + pl'. What remains is the instability when playing with the morph interface.
Han
RedHat is quite conservative with the C compiler (they want to be able to compile the kernel <g>). gcc-2.7.2.3-14. (For C++ they use egcs).
Thank you for this information. I thought that egcs was being more widely deployed than that, even to the level of compiling the prod- uction glibc. I still have bitter memories of hand hacking assembly for one platform because the vendor supplied compiler generated buggy code when the optimizer was used. While I think the bazaar approach, ostensibly the means of egcs development, is a nifty approach to some problems, I worry when it's applied to things as essential as a com- piler. I'm not left with the feeling that the bazaar is a place where proofs get done that often.
squeak-dev@lists.squeakfoundation.org
-
Christopher Oliver -
Fred Bacon -
Han Holl