Hi all-- I've reached a new minimum snapshot size (167,224 bytes uncompressed, the previous one from 2003 was 211,504 bytes uncompressed). That's for a system which supports remote browsing. I've also conceived a new design for the system tracer, one that implements it as a feature of the simulator (which didn't exist when the current tracer was written). With this tracer one will be able to write new snapshots without having to trace the object memory in which the tracer is running, and with certainty that every object in the result was required for an interpreter to function. I suspect that the smallest "graceful" object memory (e.g., one that just quits) will be something like 1000 bytes. Using Spoon's object memory visualization tools, I've noticed that even the "3 plus 4" snapshot from the Fenix project (15,192 bytes uncompressed) has a lot of unnecessary stuff in it (e.g., Characters). thanks, -C http://netjam.org/spoon -- Craig Latta improvisational musical informaticist www.netjam.org Smalltalkers do: [:it | All with: Class, (And love: it)] |
By the way I've found that it's impossible to get ANY version of the system
tracer. The original system tracer isn't available, on SqueakMap and the url that's supposed to be hosting SystemTracer2 and related utilities (LargeCollections) no longer exists. I also tried looking on the Wiki for the version of SystemTracer that the group doing the conversion to 64 bits uses. I didn't find that either. But perhaps if I had been willing to download their whole tar I would have found it. It is kind of a shame that there's no central repository that SqueakMap uses rather than trusting a bunch of urls that apparently belong to random people and companies. Josh Scholar ----- Original Message ----- From: "Craig Latta" <[hidden email]> To: <[hidden email]> Sent: Sunday, January 15, 2006 7:50 PM Subject: Spoon progress 15 January 2006: new minimum, new system tracer > > Hi all-- > > I've reached a new minimum snapshot size (167,224 bytes uncompressed, > the previous one from 2003 was 211,504 bytes uncompressed). > > That's for a system which supports remote browsing. I've also conceived > a new design for the system tracer, one that implements it as a feature > of the simulator (which didn't exist when the current tracer was > written). With this tracer one will be able to write new snapshots > without having to trace the object memory in which the tracer is > running, and with certainty that every object in the result was required > for an interpreter to function. > > I suspect that the smallest "graceful" object memory (e.g., one that > just quits) will be something like 1000 bytes. Using Spoon's object > memory visualization tools, I've noticed that even the "3 plus 4" > snapshot from the Fenix project (15,192 bytes uncompressed) has a lot of > unnecessary stuff in it (e.g., Characters). > > > thanks, > > -C > > http://netjam.org/spoon > > -- > Craig Latta > improvisational musical informaticist > www.netjam.org > Smalltalkers do: [:it | All with: Class, (And love: it)] > > > > |
On 15-Jan-06, at 9:12 PM, <[hidden email]> <[hidden email]> wrote: > By the way I've found that it's impossible to get ANY version of > the system > tracer. > > The original system tracer isn't available, Yes it is, its SM entry just hasn't been updated since I had to move my site. You wouldn't really want the old version though. I just removed it completely to remove confusion. > on SqueakMap and the url that's > supposed to be hosting SystemTracer2 and related utilities > (LargeCollections) no longer exists. Looks like the domain has lapsed. I haven't heard from Stephen Pair in ages so perhaps he's moved on to other interests. I have put a copy of it on SM as 'SystemTracerV2' to return it to accessibility. > I also tried looking on the Wiki for the version of SystemTracer > that the > group doing the conversion to 64 bits uses. I didn't find that > either. But > perhaps if I had been willing to download their whole tar I would > have found > it. > > It is kind of a shame that there's no central repository that > SqueakMap uses > rather than trusting a bunch of urls that apparently belong to random > people and companies. tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim Strange OpCodes: IXM: Initiate X-rated error Messages |
In reply to this post by joshscholar
> By the way I've found that it's impossible to get ANY version of the > system tracer. I'd expect to find it in the VMMaker package on SqueakMap, along with the rest of the platform-independent virtual machine implementation. -C -- Craig Latta improvisational musical informaticist www.netjam.org Smalltalkers do: [:it | All with: Class, (And love: it)] |
You made me to look :)
I didn't find anything called SystemTracer in any of the VMMaker packages. Joshua Scholar ----- Original Message ----- From: "Craig Latta" <[hidden email]> To: <[hidden email]> Sent: Sunday, January 15, 2006 10:16 PM Subject: re: the old system tracer (was "Spoon progress 15 January 2006: new minimum, new system tracer") > > > By the way I've found that it's impossible to get ANY version of the > > system tracer. > > I'd expect to find it in the VMMaker package on SqueakMap, along with > the rest of the platform-independent virtual machine implementation. > > > -C > > -- > Craig Latta > improvisational musical informaticist > www.netjam.org > Smalltalkers do: [:it | All with: Class, (And love: it)] > > > > |
In reply to this post by joshscholar
[hidden email] puso en su mail :
> By the way I've found that it's impossible to get ANY version of the system > tracer. > > The original system tracer isn't available, on SqueakMap and the url that's > supposed to be hosting SystemTracer2 and related utilities > (LargeCollections) no longer exists. > > I also tried looking on the Wiki for the version of SystemTracer that the > group doing the conversion to 64 bits uses. I didn't find that either. But > perhaps if I had been willing to download their whole tar I would have found > it. > > It is kind of a shame that there's no central repository that SqueakMap uses > rather than trusting a bunch of urls that apparently belong to random > people and companies. > > Josh Scholar SystemTracer2.sar (15K) Download Attachment |
In reply to this post by joshscholar
This is called squeaksource.
On 16 janv. 06, at 06:12, <[hidden email]> <[hidden email]> wrote: > By the way I've found that it's impossible to get ANY version of > the system > tracer. > > The original system tracer isn't available, on SqueakMap and the > url that's > supposed to be hosting SystemTracer2 and related utilities > (LargeCollections) no longer exists. > > I also tried looking on the Wiki for the version of SystemTracer > that the > group doing the conversion to 64 bits uses. I didn't find that > either. But > perhaps if I had been willing to download their whole tar I would > have found > it. > > It is kind of a shame that there's no central repository that > SqueakMap uses > rather than trusting a bunch of urls that apparently belong to random > people and companies. > > Josh Scholar > > ----- Original Message ----- > From: "Craig Latta" <[hidden email]> > To: <[hidden email]> > Sent: Sunday, January 15, 2006 7:50 PM > Subject: Spoon progress 15 January 2006: new minimum, new system > tracer > > >> >> Hi all-- >> >> I've reached a new minimum snapshot size (167,224 bytes uncompressed, >> the previous one from 2003 was 211,504 bytes uncompressed). >> >> That's for a system which supports remote browsing. I've also >> conceived >> a new design for the system tracer, one that implements it as a >> feature >> of the simulator (which didn't exist when the current tracer was >> written). With this tracer one will be able to write new snapshots >> without having to trace the object memory in which the tracer is >> running, and with certainty that every object in the result was >> required >> for an interpreter to function. >> >> I suspect that the smallest "graceful" object memory (e.g., one that >> just quits) will be something like 1000 bytes. Using Spoon's object >> memory visualization tools, I've noticed that even the "3 plus 4" >> snapshot from the Fenix project (15,192 bytes uncompressed) has a >> lot of >> unnecessary stuff in it (e.g., Characters). >> >> >> thanks, >> >> -C >> >> http://netjam.org/spoon >> >> -- >> Craig Latta >> improvisational musical informaticist >> www.netjam.org >> Smalltalkers do: [:it | All with: Class, (And love: it)] >> >> >> >> > > |
In reply to this post by timrowledge
On Sun, Jan 15, 2006 at 10:13:38PM -0800, tim Rowledge wrote:
> > on SqueakMap and the url that's > > supposed to be hosting SystemTracer2 and related utilities > > (LargeCollections) no longer exists. > Looks like the domain has lapsed. I haven't heard from Stephen Pair > in ages so perhaps he's moved on to other interests. > I have put a copy of it on SM as 'SystemTracerV2' to return it to > accessibility. Speaking of SystemTracer, I have a couple of questions: 1) Dan's 64 bit system tracer is tucked away at: http://squeakvm.org/squeak64/dist3/ChangeSets/System-Tracing.2.cs Could you merge this into SystemTracerV2 so it does not get lost? 2) This system tracer apparently works only for big endian machines. Can anyone offer some tips as to what needs to be changed to make it work on little enders? I know that the image header needs to be written differently, but <dumb question> are there other issues in tracing the image itself? </dumb question> Thanks, Dave |
In reply to this post by ccrraaiigg
Hi again-- I've written enough of the new tracer to write an object memory which adds 3 and 4, then quits. The snapshot[1] is 1,772 bytes uncompressed. Without using the simulator or visualization tools, about the most interesting thing you can do with it is watch it not crash. :) But I do think this is a fun proof-of-concept. -C [1] http://netjam.org/spoon/smallest.image -- Craig Latta improvisational musical informaticist www.netjam.org Smalltalkers do: [:it | All with: Class, (And love: it)] |
On 16-Jan-06, at 10:26 AM, Craig Latta wrote: > > Hi again-- > > I've written enough of the new tracer to write an object memory > which adds 3 and 4, then quits. The snapshot[1] is 1,772 bytes > uncompressed. Without using the simulator or visualization tools, > about the most interesting thing you can do with it is watch it not > crash. :) But I do think this is a fun proof-of-concept. Bloody good show! Suggestion for a Neat Demo; extend this a little to a) print out the answer as the return code or to stdio or whatever. There are some prims that should help do that without too much extra image code. I think. Maybe. Easy to add a primPrintStringToTerminal anyway b) accept simple equations as startup args Making a little Smalltalk commandline caculator app. Won't be 1700 bytes but surely less than 16kb. tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim Adding manpower to a late software project makes it later. |
On Mon, Jan 16, 2006 at 10:39:58AM -0800, tim Rowledge wrote:
> On 16-Jan-06, at 10:26 AM, Craig Latta wrote: > > I've written enough of the new tracer to write an object memory > > which adds 3 and 4, then quits. The snapshot[1] is 1,772 bytes > > uncompressed. Without using the simulator or visualization tools, > > about the most interesting thing you can do with it is watch it not > > crash. :) But I do think this is a fun proof-of-concept. > Bloody good show! Suggestion for a Neat Demo; extend this a little to > a) print out the answer as the return code or to stdio or whatever. > There are some prims that should help do that without too much extra > image code. I think. Maybe. Easy to add a primPrintStringToTerminal > anyway > b) accept simple equations as startup args > Making a little Smalltalk commandline caculator app. Won't be 1700 > bytes but surely less than 16kb. Wouldn't you know it. Someone finally gets the image trimmed down to a reasonable size, and the first thing he gets is a bunch of new feature requests ;-) Craig, this is really good. Your 1772 byte image starts right up, opens a window, and exits cleanly. Quite remarkable. Dave |
In reply to this post by David T. Lewis
On 16-Jan-06, at 7:29 AM, David T. Lewis wrote: > > Speaking of SystemTracer, I have a couple of questions: > > 1) Dan's 64 bit system tracer is tucked away at: > http://squeakvm.org/squeak64/dist3/ChangeSets/System-Tracing.2.cs > > Could you merge this into SystemTracerV2 so it does not get lost? I don't recall ever hearing of this version before, which is a bit worrying. However it has nothing in it different to the .1 file except the SystemDictionary>wordSize method that has to go into the mainstream image anyway. So no changes needed I think. > > 2) This system tracer apparently works only for big endian machines. > Can anyone offer some tips as to what needs to be changed to make > it work on little enders? I know that the image header needs to > be written differently, but <dumb question> are there other issues > in tracing the image itself? </dumb question> Don't forget that the endianness is converted on startup by the vm. Or at least, it supposed to be. Anything that isn't would be a problem. tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim "#define QUESTION ((bb) || !(bb)) - Shakespeare." |
In reply to this post by David T. Lewis
On 16-Jan-06, at 11:03 AM, David T. Lewis wrote: > > Wouldn't you know it. Someone finally gets the image trimmed > down to a reasonable size, and the first thing he gets is a > bunch of new feature requests ;-) Well of course. And we want a full reference manual and tutorial Right Now too. > > Craig, this is really good. Your 1772 byte image starts right > up, opens a window, and exits cleanly. Quite remarkable. Ooh, it opens a window? Bad VM, naughty VM, go to bed with no supper. RISC OS Squeak gets it right and doesn't open a window until there is a display to the screen. Smug grin. tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim Hardware: The parts of a computer system that can be kicked. |
On Mon, Jan 16, 2006 at 11:14:58AM -0800, tim Rowledge wrote:
> > On 16-Jan-06, at 11:03 AM, David T. Lewis wrote: > > > > Craig, this is really good. Your 1772 byte image starts right > > up, opens a window, and exits cleanly. Quite remarkable. > Ooh, it opens a window? Bad VM, naughty VM, go to bed with no supper. > RISC OS Squeak gets it right and doesn't open a window until there is > a display to the screen. Smug grin. Yer honour, if it may please the court, I submit for the record a copy of the the change set I posted in 1999 to address this very issue. Dave p.s. Based on a suggestion by a Mr. Tim Rowledge if my memory serves me well. LazyDisplayInit.15Nov515pm.cs.gz (2K) Download Attachment |
Free forum by Nabble | Edit this page |