[squeak-dev] Re: [Vm-dev] Bug in writing compressed stream when saving an mcz (was: New Cog VMs available...)

[Eliot Miranda]

On Sat, Jul 19, 2014 at 9:50 AM, tim Rowledge <tim at rowledge.org> wrote:

>
>
>
>
>>
> It is internal in the Cog VMs.  And before you condemn that choice it /is/
> slightly more efficient as it doesn't have to indirect to access the
> interpreterProxy API.  But in any case new Cog VMs are already available.
>
>
>  I know - your more direct macro access is a good idea. But that wasnt
> the point I intended; unless we've broken the plugin mechanism over the
> years a fixed-up zip plugin will over-ride the broken internal code,
> offering a very easy fix.
>


Oh *cool*!  So the loader prefers shared objects/dlls over internals if
available?  Great idea.  However, it does seem to me that building and
deploying a new VM is about the same complexity as building and deploying a
new plugin.  In fact, slightly less, cuz installing the plugin in an
existing VM is, for some, tricky.


> It should even work in a running system where you don't want to quit and
> restart with a new vm for whatever reason. Something like
> Smalltalk unloadModule: 'ziplthingy'
> Should disconnect from the old version and the next call to anything in
> the plugin ought to load the new external copy. Obviously this could be
> complicated by any stored state, so caveat emptor.
>
> It's a really neat trick when developing a plugin.
>

 Yes indeed.

-- 
aloha,
Eliot
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/vm-dev/attachments/20140719/ed1e24b4/attachment.htm