[squeak-dev] Re: New CogVMs available
Eliot Miranda
eliot.miranda at gmail.com
Sat Sep 6 14:13:01 UTC 2014
Hi Frank,
On Sep 5, 2014, at 2:06 PM, Frank Shearar <frank.shearar at gmail.com> wrote:
> On 5 August 2014 17:43, Eliot Miranda <eliot.miranda at gmail.com> wrote:
>>
>>
>>
>> On Tue, Aug 5, 2014 at 8:13 AM, Frank Shearar <frank.shearar at gmail.com>
>> wrote:
>>>
>>> On 5 August 2014 16:00, Eliot Miranda <eliot.miranda at gmail.com> wrote:
>>>> Hi Frank,
>>>>
>>>> On Aug 5, 2014, at 2:20 AM, Frank Shearar <frank.shearar at gmail.com>
>>>> wrote:
>>>>
>>>>> On 5 August 2014 03:49, Eliot Miranda <eliot.miranda at gmail.com> wrote:
>>>>>> ...at http://www.mirandabanda.org/files/Cog/VM/VM.r3058/.
>>>>>>
>>>>>> CogVM binaries as per VMMaker.oscog-eem.842/r3058
>>>>>
>>>>> I'll update CI this evening.
>>>>>
>>>>> Quick question - for the Linux VMs, am I correct in thinking that MT
>>>>> VMs are _always_ HT? (And that on Windows they are never?)
>>>>
>>>> Mac and Win VMs have only heartbeat threads. Linux VMs have either a
>>>> heartbeat thread or the ITIMER heartbeat. If the linux archive contains
>>>> "ht" it's a threaded heartbeat VM, period. The naming is screwed up because
>>>> ITIMER used to be the only option on linux. Would changing the names so
>>>> that linux ITIMER VMs were singled out be useful?
>>>
>>> Mainly I'm trying to figure out how to map "kind of VM" to a file
>>> name. I just noticed that
>>> http://www.mirandabanda.org/files/Cog/VM/VM.r3058/ has a cogmtlinuxht,
>>> but not a cogmtlinux, VM. But there are both coglinux and coglinuxht
>>> and images, so it looks like MT and HT are orthogonal switches (on
>>> Linux, at least).
>>
>>
>> Yes. And there is no such switch elsewhere. And ITIMER will eventually die
>> the death.
>
> With 3063 has ITIMER died the death? (I ask because there's no
> coglinux VM, only coglinuxht.)
There is a Cog ITIMER VM but no Spur ITIMER VM. I can add one if requested but hope that it won't be needed.
> frank
>
>>> If they are, that's great, and I need to extend my
>>> existing kind-of-VM to reflect this, because at the moment it knows
>>> about :mt, :spur and :normal VMs, whereas it might need to know about
>>> :mtht, :mt and :ht VMs. Make sense?
>>
>>
>> Yes.
>>
>>>
>>> (https://github.com/squeak-smalltalk/squeak-ci/blob/master/lib/squeak-ci/cog.rb
>>> has my horrible hacks: look at cog_location and filename for the
>>> current mapping of kind-of-VM -> string.
>>>
>>> frank
>>
>> --
>> best,
>> Eliot
More information about the Squeak-dev
mailing list
|