Smalltalk › Pharo › Pharo Smalltalk Developers

Re: New Latest VM - Please Test

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

1 message

Boris Shingarov-2

Re: New Latest VM - Please Test

...hmm, more interesting facts upon investigation. Doesn't look like the vanilla VMMaker.image that gets built by "cmake . ; make" upon virgin clone, is very healthy. For example, when loading VMMaker, we try to extend ContextPart, and that class does not exist, so loading BytecodeSets couldn't have succeeded. Is this supposed to be part of the VMMaker development image? How can one tell? The "make" reports the build was successful, so presumably VMMaker.image must be good?

-----Boris Shingarov/Employee/LW-US wrote: -----

To: [hidden email]
From: Boris Shingarov/Employee/LW-US
Date: 02/21/2020 05:34AM
Cc: [hidden email]
Subject: [Pharo-dev] New Latest VM - Please Test

Dear Pablo and other friends:

Today I finally got around to actually getting my hands on the 'headless' codebase, and I must admit I feel rather confused (if not even somewhat stupid). I built the VM and went to use the generated build/vmmaker/VMMaker.image to look around (step through tests etc.) I thought I would start with what I had always regarded as the simplest tests, the Cog*CompilerTests. Well, in the "build-process-generated" VMMaker image these don't even compile, e.g. CogIA32CompilerTests>>processor references BochsIA32Alien which is not present in the image. Then, how does one simulate the Pharo-VM at all, without the aliens? Am I missing something obvious? I'd think the main README.md would explain if the development process is different from the Cog branch (iow, from the Squeak-based process), but it seems silent on that topic. Perhaps you can explain?

Best,
Boris