Iterating through all image objects

Brian Rice water at tunes.org
Thu Nov 11 08:19:18 PST 2004


I forgot to mention: a VM bug-fix which Lee applied led him to 
eliminate a header bit (the one indicating free object areas) in 
exchange for more identity-hash value space. So there is a fundamental 
image-format update here.

On Nov 11, 2004, at 8:11 AM, Brian Rice wrote:

> Bug fixes and new alpha images are up. Enjoy!
>
> On Nov 10, 2004, at 12:34 PM, Brian Rice wrote:
>
>> I've applied more fixes now to the userland for the 
>> delegates/referrers streams. They work well, but be warned when using 
>> this that any objects that come through may have issues with being 
>> printed on the REPL. So errors will commonly be due to that.
>>
>> On Nov 10, 2004, at 12:17 PM, Brian Rice wrote:
>>
>>> These bugs are now fixed in CVS, due to one issue on the VM side and 
>>> a couple of problems in my user-land code. I have not verified 
>>> referrer-streams correct operation yet, however, but the others 
>>> behave correctly and safely; just be aware that going through the 
>>> elements will expose you to all sorts of objects.
>>>
>>> A new alpha set should be uploaded sometime today. Lee is working on 
>>> some minor VM format changes.
>>>
>>> On Nov 10, 2004, at 9:21 AM, Brian Rice wrote:
>>>
>>>> This feature that I mentioned is now implemented in a simple way in 
>>>> the cvs code tree, but alpha VMs/images have not been uploaded as 
>>>> using it is very crash-prone. Userland code for it is in 
>>>> src/mobius/vm/interp/memory.slate and provides an "Image reader" 
>>>> iterator as well as a #childStream method for all objects. There's 
>>>> also a tentative stream for objects referring to a target object.
>>>>
>>>> Including this feature in builds should not cause issues if you are 
>>>> not using it, but be aware that using it will eventually cause bad 
>>>> crashes for now.
>>>>
>>>> On Nov 7, 2004, at 3:30 PM, Brian Rice wrote:
>>>>
>>>>> No, not yet. I need to expose the all-object, all-referring-to, 
>>>>> and all-delegating-to iteration protocol (see memory.slate) in 
>>>>> some primitives to build a toolkit for this. It's not too hard, 
>>>>> but I hadn't finished it yet. The idea is that on top of these 
>>>>> primitives you would be able to get streams to traverse the Image 
>>>>> and build on that.
>>>>>
>>>>> I'll inform when it's ready (maybe I'll push and make that today).
>>>>>
>>>>> On Nov 7, 2004, at 2:50 PM, Todd Fleming wrote:
>>>>>
>>>>>> Let's say I have a prototype P with some slots. Let's also say 
>>>>>> that P traits has several roles. I clone P several times. I 
>>>>>> discover that I need to add some more slots to P and some more 
>>>>>> roles to P traits that need those slots. Is there an easy way to 
>>>>>> add the slots to existing objects that delegate to P traits?
>>>>
>>>> --
>>>> Brian T. Rice
>>>> LOGOS Research and Development
>>>> http://tunes.org/~water/
>>>>
>>>>
>>> --
>>> Brian T. Rice
>>> LOGOS Research and Development
>>> http://tunes.org/~water/
>>>
>>>
>> --
>> Brian T. Rice
>> LOGOS Research and Development
>> http://tunes.org/~water/
>>
>>
> --
> Brian T. Rice
> LOGOS Research and Development
> http://tunes.org/~water/
>
>
--
Brian T. Rice
LOGOS Research and Development
http://tunes.org/~water/




More information about the Slate mailing list