[squeak-dev] Re: [Vm-dev] Bug in writing compressed stream when
saving an mcz (was: New Cog VMs available...)
eliot.miranda at gmail.com
Mon Jul 21 09:47:23 UTC 2014
On Mon, Jul 21, 2014 at 2:32 AM, Bert Freudenberg <bert at freudenbergs.de>
> On 19.07.2014, at 02:15, Eliot Miranda <eliot.miranda at gmail.com> wrote:
> On Fri, Jul 18, 2014 at 4:20 PM, David T. Lewis <lewis at mail.msen.com>
>> On Fri, Jul 18, 2014 at 02:06:18PM -0700, Eliot Miranda wrote:
>> > I think we have to accept that the fix to on:from:to: means that exsting
>> > VMs are broken. Certainly the ZipPlugin (InflatePlugin &
>> > dependency on WriteStream and ReadStream inst size is restrictive. I'd
>> > rather lift the restriction and use new VMs than live with the
>> > long-term.
>> Is this fix is of sufficient urgency to justify asking all Squeak trunk
>> to replace their VMs? Is there any less disruptive way to address the
>> I don't know the answers, just asking.
> Well, I've fixed the VM (commit to appear shortly) and much prefer having
> a fixed plugin. The alternatives are
> a) having some horrible hack in the image to store initialPositionOrNil
> out of streams
> b) living with on:from:to: being broken
> Since VMs can be fixed it seems to me that the best solution is to build
> new VMs. That's the solution I'm following anyway.
> But ... breaking old images is *not* a good idea.
> IMHO we should name the fixed plugin ZipPlugin2. The new image with
> changed ivar offsets will use that plugin. Old images with the original
> object layout will use the original.
But old images will continue to work with either plugin. And any image can
back out of the fix to on:from:to: and delete the added inst vars and
continue to work. What breaks things is adding an inst var to WriteStream
or ReadStream, /not/ changing the plugin.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Vm-dev