V3
Locked
 
|
Reading the code I again noticed "Squeak V3" as name for the old bytecode set. Always wanted to know: Where does the "V3" nomenclature come from? What's V1 and V2? And should we consider Sista to be V4, or treat it as completely unrelated? Vanessa |
|
Vanessa,
> Reading the code I again noticed "Squeak V3" as name for the old bytecode > set. Always wanted to know: Where does the "V3" nomenclature come from? It was the format for the planned Squeak 3.0: https://wiki.squeak.org/squeak/750 > What's V1 and V2? They were never named, but you could consider the original Squeak to be V1 and the one with Morphic in it to be V2, but I don't think there are any significant changes between them. But please note that the V3 effort was abandoned, so in practice VI3 is essentially the original Squeak image format. > And should we consider Sista to be V4, or treat it as completely unrelated? There was a VI4 effort, also abandoned, that is linked from that page: https://wiki.squeak.org/squeak/2119 But Eliot's efforts to add closures did get adopted (third time's the charm!) and yet it only got new image version numbers and not a new name. So it is a VI3 format that doesn't match the previous meaning of the term. Since none of those matter any more it is just odd but doesn't actually cause any confusion. What is confusing is the use of the term both for the image format (VI3 x Spur) and the bytecode set (VI3 x Sista). In any case, the 3 in VI3 was originally associated with the transition from Squeak 2 to Squeak 3 and makes no sense in this Squeak 5 (soon 6) era. -- Jecel |
|
|
So with regard to the VM and in particular the byte code set, “V3” is completely wrong? Should we rename the Encoder subclass and related methods to something reasonable then? I believe Eliot introduced these names, what do you think? Vanessa On Tue, Jan 5, 2021 at 7:03 PM Jecel Assumpcao Jr <[hidden email]> wrote: Vanessa, |
|
|
In reply to this post by codefrau
Hi V, On Tue, Jan 5, 2021 at 6:07 PM Vanessa Freudenberg <[hidden email]> wrote:
In my diseased and uninformed mind V3 came from my noticing that the double extended do anything bytecode didn't exist in Smalltalk-80, so I presumed it wasn't in Squeak V1. I don't know if it's in V2, and I was too lazy to go look, but I did know it was in V3, so I called it, and the object representation "V3". Presumptuous but concise of me. For me it really is a label for no more than the state of play of Squeak when I came to start changing its core execution engine. I hope this is harmless enough. _,,,^..^,,,_ best, Eliot |
|
|
On Fri, Jan 08, 2021 at 08:08:33PM -0800, Eliot Miranda wrote:
> Hi V, > > On Tue, Jan 5, 2021 at 6:07 PM Vanessa Freudenberg <[hidden email]> > wrote: > > > Reading the code I again noticed "Squeak V3" as name for the old bytecode > > set. Always wanted to know: Where does the "V3" nomenclature come from? > > What's V1 and V2? And should we consider Sista to be V4, or treat it as > > completely unrelated? > > > > In my diseased and uninformed mind V3 came from my noticing that the double > extended do anything bytecode didn't exist in Smalltalk-80, so I presumed > it wasn't in Squeak V1. I don't know if it's in V2, and I was too lazy to > go look, but I did know it was in V3, so I called it, and the object > representation "V3". Presumptuous but concise of me. For me it really is > a label for no more than the state of play of Squeak when I came to start > changing its core execution engine. I hope this is harmless enough. > I have become quite comfortable with the "V3" notation by now, regardless of its origins. I would prefer not to change it because I have used it extensively in some of my own personal projects (e.g. all of the package names in my http://www.squeaksource.com/TrunkUpdateStreamV3 hobby project). I also just like the name. "V3" is short and easily recognized, and it works nicely as a prefix or suffix in other names. To me, it also carries a vague connotation of "stable but not necessarily the latest thing". That's probably based on my recollection of early PC applications and operating systems where everyone knew not to trust any product until it reached at least version 3 ;-) Dave |
|
|
On Sat, Jan 9, 2021 at 7:09 AM David T. Lewis <[hidden email]> wrote: On Fri, Jan 08, 2021 at 08:08:33PM -0800, Eliot Miranda wrote: Can we come up with some major VM milestones to retroactively label as V1 / V2 / V3? Off the top of my head these could be * the original extension from ST80 to Squeak (as documented in Dan's "Back to the future": 32 bit direct pointers, no refcounting, variable-length object headers, etc) * named primitives * 64 bit support * closures V1 would be the 1996 one then V3 might be the closure-supporting one (which year was that?) V2 - ? Vanessa |
|
|
On Mon, Jan 11, 2021 at 2:57 PM Vanessa Freudenberg <[hidden email]> wrote:
+1
2009 (internally @ Qwaq) / 2010 (released to the community)
Named primitives seems like a major change to me. _,,,^..^,,,_ best, Eliot |
|
|
On Mon, Jan 11, 2021 at 3:00 PM Eliot Miranda <[hidden email]> wrote:
This message appears to indicate that named primitives and weak collections arrived at the same time in 1998: Seems like a good candidate. Unless someone remembers a history of changes to bytecodes specifically? - Vanessa - |
|
|
In reply to this post by Eliot Miranda-2
> On 2021-01-11, at 3:00 PM, Eliot Miranda <[hidden email]> wrote: > > V1 would be the 1996 one then > > +1 > V3 might be the closure-supporting one (which year was that?) Except Dan was talking V3 back in '99 and the intent was to include Ian's never-quite-done jit, and from a Dan-mail of Oct 19th - > My revised goal would be... > > Increase max context size to 80, and go back to two sizes like 16 and 80 > Merge format bits in behaviors > Merge prim index bits in methods > Your compiled method changes (to which Ian might wish to add an extra field) > > 2009 (internally @ Qwaq) / 2010 (released to the community) > > V2 - ? > > Named primitives seems like a major change to me. The named/external prims was early '99-ish tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim Oyster (n.), a person who sprinkles his conversation with Yiddishisms. |
|
|
On Mon, Jan 11, 2021 at 3:15 PM tim Rowledge <[hidden email]> wrote:
Do you mean "3.x"? That was in the Squeak 2.x era and I think he was talking about Squeak versions. We're trying to find a justification for calling the "old" VMs "V3", independent of the Squeak releases (or SqueakV3.sources which unfortunately uses V3 in a completely different sense) I'd personally prefer to get rid of the V3 naming for bytecode/VM-related entities but as Dave pointed out, it has proliferated due to being unquestioned for too long. Giving some sensible meaning to "V3" seems like a good way forward. and the intent was to include Ian's never-quite-done jit, and from a Dan-mail of Oct 19th - Right, but that version never got adopted. The named/external prims was early '99-ish Sounds about right. Does someone have an actual history of the VM, maybe from SVN? Or were we still using changesets for VM code back then? - Vanessa - |
|
|
> On 2021-01-11, at 3:37 PM, Vanessa Freudenberg <[hidden email]> wrote: > > On Mon, Jan 11, 2021 at 3:15 PM tim Rowledge <[hidden email]> wrote: > > > > On 2021-01-11, at 3:00 PM, Eliot Miranda <[hidden email]> wrote: > > > > > V1 would be the 1996 one then > > > > +1 > > V3 might be the closure-supporting one (which year was that?) > > Except Dan was talking V3 back in '99 > > Do you mean "3.x"? That was in the Squeak 2.x era and I think he was talking about Squeak versions. Maaaaaybe. The emails are headed 'Version 3 image'. I think retconning history may be futile. It's quite clear the reality lines have been thoroughly messed up since at least 2000. tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim "How many Motie Mediators does it take to chage a lightbulb?” "Are you insane? Only Crazy Eddie would want to change *anything*!" |
|
|
In reply to this post by codefrau
On Mon, Jan 11, 2021 at 03:37:07PM -0800, Vanessa Freudenberg wrote:
> On Mon, Jan 11, 2021 at 3:15 PM tim Rowledge <[hidden email]> wrote: > > > The named/external prims was early '99-ish > > > > Sounds about right. Does someone have an actual history of the VM, maybe > from SVN? Or were we still using changesets for VM code back then? > The SVN repository history is on line, and begins October 2001: http://squeakvm.org/cgi-bin/viewvc.cgi/squeak/trunk/platforms/Cross/vm/sqVirtualMachine.h?revision=2&view=markup Prior to that the support code was hosted on sourceforge.com. I don't know if any copy of that repository still exists, and I don't know when it first became publicly available. Some of that history may be lost. The Smalltalk sources were first put on squeaksource.org in 2004 (MMMaker-tpr.2). As far as I know, the main gap in this historical record would probably be the sourceforge history. Dave |
|
|
David T. Lewis wrote on Mon, 11 Jan 2021 20:04:32 -0500
> The SVN repository history is on line, and begins October 2001: > > http://squeakvm.org/cgi-bin/viewvc.cgi/squeak/trunk/platforms/Cross/vm/sqVirtualMachine.h?revision=2&view=markup The VM was originally found in the .sources and .changes files, so looking at the #initializeBytecodeTable method I can see that the only change was that Squeak 1.1, 1.16 and 1.18 had (138 143 unknownBytecode) and that got replaced in Squeak 1.22 by (138 143 experimentalBytecode) There were no more changes until at least Squeak 3.8. Eventually the code was moved to VMMaker. For the history of named primitives and similar things I would first look into the code in the compiler that parses the methods. -- Jecel |
|
|
In reply to this post by David T. Lewis
Hi Dave-- > ... > > Prior to that the support code was hosted on sourceforge.com. I don't > know if any copy of that repository still exists, and I don't know > when it first became publicly available. Some of that history may be > lost. Perhaps it's worth looking around the Wayback Machine snapshots at archive.org? -C -- Craig Latta :: research computer scientist Black Page Digital :: Berkeley, California 663137D7940BF5C0AFC :: 1349FB2ADA32C4D5314CE |
|
|
Hi Craig,
On Thu, Jan 14, 2021 at 01:35:42PM -0800, Craig Latta wrote: > > Hi Dave-- > > > ... > > > > Prior to that the support code was hosted on sourceforge.com. I don't > > know if any copy of that repository still exists, and I don't know > > when it first became publicly available. Some of that history may be > > lost. > > Perhaps it's worth looking around the Wayback Machine snapshots at > archive.org? > I don't know if there is anything of real interest in the original SourceForge repository, but if so I would guess that Ian Piumarta would be most likely to have a copy of it. The current Subversion repository that is still on line at squeakvm.org has version history going back to October 2001. Most likely there is some discussion of this on the squeak-dev list archives from that period. I don't recall any specifics, but I think that sourceforge was the github of that era - ie the latest and greatest cool hosting service - and for reasons I no longer recall it got to be uncool and we had to rescue the VM sources. "We" in this case being Ian, who has been providing and paying for the hosting service ever since. FWIW, Tim Rowledge appears to have the honor of having made the very first commit to the SVN repository circa 2001. I keep a private git mirror of the squeakvm.org repository, and a tail of the git log says this: > commit 6364e4a829c239562a1dbddbf77c57fe74c20a3c > Author: Tim Rowledge <[hidden email]> > Date: Wed Oct 24 23:11:49 2001 +0000 > > Initial revision > > commit eaae7a667f99ee68cc477d3167a6fe3b88e012bf > Author: (no author) <(no author)> > Date: Wed Oct 24 23:11:49 2001 +0000 > > New repository initialized by cvs2svn. Which may or may not be correct, since the actual SVN log says: > ------------------------------------------------------------------------ > r2 | rowledge | 2001-10-24 19:11:49 -0400 (Wed, 24 Oct 2001) | 2 lines > > Initial revision > > ------------------------------------------------------------------------ Either way I think that Tim gets first commit honors. And the git variant of the log tells me that the earlier repo was a CVS repository, for whatever that may be worth. Dave |
|
|
> On 2021-01-14, at 8:07 PM, David T. Lewis <[hidden email]> wrote: > > I don't recall any specifics, but I think that sourceforge was the github > of that era - ie the latest and greatest cool hosting service - and for > reasons I no longer recall it got to be uncool and we had to rescue the > VM sources. "We" in this case being Ian, who has been providing and paying > for the hosting service ever since. I vaguely recall us getting uncomfortable with the ethos of sourceforge for some reason. Were they bought up by some nasty bunch? Started supporting the wrong feetball team? Got bossy about precise licensing terms? > > FWIW, Tim Rowledge appears to have the honor of having made the very first > commit to the SVN repository circa 2001. Uuuh, gosh. My emails imply we made the move around this time on '05. Ah - here is the announcement email Hi all! For you who follow the SF Squeak-VM mailinglist this isn't a surprise. The source in CVS at SourceForge is not up to date and hasn't been for quite a while. But the good news is that Ian and Co. have set up a Subversion repository with the fresh sources. This SVN (Subversion) repository is located here: http://squeak.hpl.hp.com/svn/squeak/trunk/platforms And if you install Subversion you should be able to just: svn co http://squeak.hpl.hp.com/svn/squeak/trunk/platforms ...to check it out anonymously. Subversion is a "CVS improved" and has lately moved to fill the shoes of CVS. The fact that the repository is located on a HP server has natural reasons - Ian Piumarta and Andreas Raab (and other VM devs?) have fast access to that server which makes it convenient for them - which is of course important. We are going to set up a mirror of this repository on the new SqF server though. We just need to learn how! :) Sidenote: If someone already knows how to do SVN mirroring, feel free to help us out. regards, Göran PS. We are going to leave SF as soon as we have all things nailed down. For example, we need to move the file downloads to the SqF server and set up one or two mirrors. tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim The gains I make don't make me a hero; all the work I do is just to get back to Zero |
|
|
On Thu, Jan 14, 2021 at 08:22:01PM -0800, tim Rowledge wrote:
> > > > On 2021-01-14, at 8:07 PM, David T. Lewis <[hidden email]> wrote: > > > > I don't recall any specifics, but I think that sourceforge was the github > > of that era - ie the latest and greatest cool hosting service - and for > > reasons I no longer recall it got to be uncool and we had to rescue the > > VM sources. "We" in this case being Ian, who has been providing and paying > > for the hosting service ever since. > > > I vaguely recall us getting uncomfortable with the ethos of sourceforge for some reason. Were they bought up by some nasty bunch? Started supporting the wrong feetball team? Got bossy about precise licensing terms? > > > > > FWIW, Tim Rowledge appears to have the honor of having made the very first > > commit to the SVN repository circa 2001. > > Uuuh, gosh. My emails imply we made the move around this time on '05. Ah - here is the announcement email > Excellent, that's it then. The earlier SourceForge CVS repository was converted to Subversion in the 2005 time frame. The conversion was done with the cvs2svn tool, which would have preserved all available version history. All of the original source history from SourceForge is present in the SVN repository on squeakvm.org now, and nothing has been lost. Dave |
|
|
Le ven. 15 janv. 2021 à 18:30, David T. Lewis <[hidden email]> a écrit :
> > On Thu, Jan 14, 2021 at 08:22:01PM -0800, tim Rowledge wrote: > > > > > > > On 2021-01-14, at 8:07 PM, David T. Lewis <[hidden email]> wrote: > > > > > > I don't recall any specifics, but I think that sourceforge was the github > > > of that era - ie the latest and greatest cool hosting service - and for > > > reasons I no longer recall it got to be uncool and we had to rescue the > > > VM sources. "We" in this case being Ian, who has been providing and paying > > > for the hosting service ever since. > > > > > > I vaguely recall us getting uncomfortable with the ethos of sourceforge for some reason. Were they bought up by some nasty bunch? Started supporting the wrong feetball team? Got bossy about precise licensing terms? > > > > > > > > FWIW, Tim Rowledge appears to have the honor of having made the very first > > > commit to the SVN repository circa 2001. > > > > Uuuh, gosh. My emails imply we made the move around this time on '05. Ah - here is the announcement email > > > > Excellent, that's it then. The earlier SourceForge CVS repository > was converted to Subversion in the 2005 time frame. The conversion > was done with the cvs2svn tool, which would have preserved all > available version history. > > All of the original source history from SourceForge is present in > the SVN repository on squeakvm.org now, and nothing has been lost. > > Dave > > Given that we still have a script named ./scripts/updateSCCSVersions I would not be amazed that some earlier history could eventually be found before the CVS time frame ;) Nicolas |
|
|
On Fri, Jan 15, 2021 at 06:40:09PM +0100, Nicolas Cellier wrote:
> Le ven. 15 janv. 2021 ?? 18:30, David T. Lewis <[hidden email]> a ??crit : > > > > On Thu, Jan 14, 2021 at 08:22:01PM -0800, tim Rowledge wrote: > > > > > > > > > > On 2021-01-14, at 8:07 PM, David T. Lewis <[hidden email]> wrote: > > > > > > > > I don't recall any specifics, but I think that sourceforge was the github > > > > of that era - ie the latest and greatest cool hosting service - and for > > > > reasons I no longer recall it got to be uncool and we had to rescue the > > > > VM sources. "We" in this case being Ian, who has been providing and paying > > > > for the hosting service ever since. > > > > > > > > > I vaguely recall us getting uncomfortable with the ethos of sourceforge for some reason. Were they bought up by some nasty bunch? Started supporting the wrong feetball team? Got bossy about precise licensing terms? > > > > > > > > > > > FWIW, Tim Rowledge appears to have the honor of having made the very first > > > > commit to the SVN repository circa 2001. > > > > > > Uuuh, gosh. My emails imply we made the move around this time on '05. Ah - here is the announcement email > > > > > > > Excellent, that's it then. The earlier SourceForge CVS repository > > was converted to Subversion in the 2005 time frame. The conversion > > was done with the cvs2svn tool, which would have preserved all > > available version history. > > > > All of the original source history from SourceForge is present in > > the SVN repository on squeakvm.org now, and nothing has been lost. > > > > Dave > > > > > > Given that we still have a script named > > ./scripts/updateSCCSVersions > > I would not be amazed that some earlier history could eventually be > found before the CVS time frame ;) > > Nicolas > LOL you are probably right! Dave |
«
Return to Squeak - Dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |