darcs patch: Support an interrupt flag, and primitive... (and 4 more)

Brian Rice water at tunes.org
Tue Nov 29 09:59:34 PST 2005


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

If you are working from the main repository, there is a patch which  
breaks new bootstraps (just unpull it to get it removed from your  
local repository), titled:

"Added comments to Root methods and replaced the use of withSeen: in  
primitive slot recursion methods with &seen:."

You do not need to bootstrap in any case to get the work-around,  
since that just modifies boot.c; running the default make target to  
build the VM should be sufficient.

However, I have not seen this error you are describing, so I cannot  
address that specifically.

Basically, don't bother until you have a good reason to (for now). :)

On Nov 29, 2005, at 7:15 AM, Márton Sasvári (IJ/ETH) wrote:

> I've tried the process described below to work around the  
> InterruptFlag issue, but 'make newboot' got into an infinite loop  
> after stepping into the debugger. The original problem is that it  
> does not find the message #parenthesizes on 'repl resource'. All  
> this on cygwin.
>
> What is the current workaround the InterruptFlag or I should not  
> bother as I've just restarted work on the json parser ?
>
> Regards,
> Marton
>
> Brian Rice wrote:
>
>> The alpha repository now has a bootstrapped set of images and the  
>> new  VM sources, and has all the patches from main applied. It is  
>> as  stable as it was before. ;)
>> On Oct 27, 2005, at 1:46 PM, Brian Rice wrote:
>>> I've turned this trick for boot.c into a patch and then  
>>> immediately  issued a rollback for it, along with an extended  
>>> comment describing  its purpose and how to bootstrap the previous  
>>> patches, so that this  can be more easily reproduced by others or  
>>> in the future if need be.
>>>
>>> I think I will use this kind of approach in the future to record   
>>> bootstrapping issues, for archival / reproducibility purposes.
>>>
>>> On Oct 27, 2005, at 6:36 AM, Tony Garnock-Jones wrote:
>>>
>>>> OK, I've tried it on my mac. I took a copy of the head of main,   
>>>> with all patches applied, and cleaned up (make distclean and   
>>>> remove the libtool detritus). Typing "make" downloaded vm.c,  
>>>> vm.h,  big.image as expected and tried to build a vm. Of course,  
>>>> the  changes in boot.c don't work with the alpha vm.c, so I  
>>>> edited  boot.c temporarily thus:
>>>>
>>>> #if 0
>>>>     if (InterruptFlag != False) {
>>>>       InterruptFlag = False;
>>>>     } else {
>>>>       break;
>>>>     }
>>>> #else
>>>>     break;
>>>> #endif
>>>>
>>>> ... and next time, "make" succeeded. Then I did "make newboot",   
>>>> which went without a hitch; then "make VM=vm.new"; then "./ 
>>>> vm.new  big.new.image", which worked too.
>>>>
>>>> Now I'm undoing the temporary hack to boot.c, and I'll see if  
>>>> the  cooperative-threading thing works like it did on Linux last  
>>>> night.
>>>>
>>>> Tony
>>>>
>>>
>>> -- 
>>> -Brian
>> -- 
>> -Brian

- --
- -Brian
http://tunes.org/~water/brice.vcf

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)

iD8DBQFDjJcMPVkAvLW1zf4RAphyAKCl0Sb/Ltz5j3+Z+QyFt6pJHw5nmgCfcFtF
iadO7YM7cmLvcQU377Asnoo=
=eya6
-----END PGP SIGNATURE-----




More information about the Slate mailing list