Falsehoods programmers believe about Smalltalk
Locked
 
|
Hi there, I just created a GitHub repo to collect myths around Smalltalk-based technologies: Pharo, Squeak, VW, VAST, Smalltalk/X, GNU/ST, etc. in the spirit of the Falsehoods lists [1-4]. This is just a draft now but please feel free to add falsehoods based on your own experiences. Examples are greatly appreciated. Cheers, Hernán |
Re: Falsehoods programmers believe about Smalltalk
|
|
Hi Hernán, On Sun, Jan 20, 2019 at 2:31 PM Hernán Morales Durand <[hidden email]> wrote:
You want pull requests? If not, would you give me write permission? I'd love to add to the "Smalltalk is obsolete" section...
_,,,^..^,,,_ best, Eliot |
|
|
Done. I have some possible myths, but I'd like to confirm or reject: - All Smalltalk bytecode sets are stack-based VM. (?) - Bytecodes are always fixed-size. (?)- Most of the time spent by a VM is in the instruction interpreter. (actually it's in the GC right?) - You cannot serialize objects containing blocks. (IIRC one can use MessageSends) - Image cannot be bootstrapped. (This is possible in ST/X and now in Pharo I think). - All Smalltalks includes UI classes. (GemStone doesn't have AFAIK). - All implementations uses direct pointers, (GST?) - All implementations uses green threads. (VAST? MT?) I'm sure people in this list will have a lot more myths heard from Conferences, Forums, Videos, Talks, etc. Like the guy who said Smalltalk was dead. So if you did something which could be ignored publicly, please don't hesitate to reply or ping me to get added as collaborator. Cheers, Hernán El dom., 20 ene. 2019 a las 22:41, Eliot Miranda (<[hidden email]>) escribió:
|
Re: Falsehoods programmers believe about Smalltalk
|
|
Hernán Morales Durand wrote on Mon, 21 Jan 2019 18:13:45 -0300
> I have some possible myths, but I'd like to confirm or reject: > > - All Smalltalk bytecode sets are stack-based VM. (?) > - Bytecodes are always fixed-size. (?) SOAR (Smalltalk On A RISC, now renamed as RISC-III) used a 32 bit register based instruction set and Smalltalk sources were translated to that. They did regret dumping bytecodes later in their project as some things became more complicated and the increase in memory use was very expensive at that time. https://apps.dtic.mil/dtic/tr/fulltext/u2/a172800.pdf SOM (Simple Object Machine) is a set of Smalltalk VMs where some of them represent code as abstract syntax trees instead of bytecodes. http://som-st.github.io/ http://www.hpi.uni-potsdam.de/hirschfeld/projects/som/ > - Most of the time spent by a VM is in the instruction interpreter. (actually it's in the GC right?) That will vary from one VM to another. On page 179 of the "green book" you can see a nice graph of the space and time used by different part of the Apple Smalltalk (from which Squeak evolved) and on page 177 you can find the numbers used to create the chart. http://sdmeta.gforge.inria.fr/FreeBooks/BitsOfHistory/ 10.2% of the time in the fetch loop, 16.0% in the bytecode interpreter, 39.2% in sends and returns, 22.6% in the memory management and 10% in primitives. Adding the first 3 numbers you get 10.2+16.0+39.2 = 65.4% in the instruction interprter while the GC is part of the 22.6% which is the memory management. That said, in Squeak gcc tricks really helped with the fetch loop and the stack VM greatly reduced the send/return overhead. So it might be the case that the GC dominates performance. Or not - we have to measure and see. > - You cannot serialize objects containing blocks. (IIRC one can use MessageSends) Given that the image contains blocks, that can't be true. Obviously serializing a subset of objects is a harder problem than just dumping memory, but I consider images a proof of existence. > - Image cannot be bootstrapped. (This is possible in ST/X and now in Pharo I think). Little Smalltalk is a good example of taking a textual representation and bootstrapping an image from it. GNU Smalltalk didn't even use images the last time I looked at it. I consider Self to be a Smalltalk (just not a Smalltalk-80) and it can start with either a snapshot (its name for image) or with an empty world and load text files (possible because the source to bytecode compiler is included in the VM). > - All Smalltalks includes UI classes. (GemStone doesn't have AFAIK) The MS-DOS port of Squeak had no GUI, just a command line prompt. That was also the case for GNU Smalltalk and Little Smalltalk. > .- All implementations uses direct pointers, (GST?) The RoarVM for Squeak uses object tables. In fact, the lack of direct pointers in early implementations is what led to the use of #become: which complicated the adoption of direct pointers. VisualWorks has an indirection pointer in the header - see slide 7 of https://www.slideshare.net/esug/spur-a-new-object-representation-for-cog > - All implementations uses green threads. (VAST? MT?) I would say this was a side effect of patching the original Smalltalk, which was its own operating system (and so the idea of green thread doesn't apply) to run on top of Unix on commercial workstations. All the old code supposed the mix of cooperative and preemptive multithreading that breaks down if you have multiple native threads. Some from scratch Smalltalks copied this model while others (I am pretty sure it was the case for MT, as you mentioned) had their libraries written with native threads in mind. -- Jecel |
|
|
Hi Jecel, El lun., 21 ene. 2019 a las 22:27, Jecel Assumpcao Jr. (<[hidden email]>) escribió: Hernán Morales Durand wrote on Mon, 21 Jan 2019 18:13:45 -0300 Yes, sadly some of the LS implementations were lost in time and need to be tracked now, http://www.littlesmalltalk.org is now chinese thing, PDST and Parla were based in LittleSmalltalk 3 but only accessible through archive: http://web.archive.org/web/20051025043437/http://www.copyleft.de/Parla/Parla.html. There is another implementation now : https://github.com/0x7CFE/llst however didn't checked yet. GNU Smalltalk didn't even use images I always wondered about how much performance is gained moving the all the Compiler infrastructure into the VM. > - All Smalltalks includes UI classes. (GemStone doesn't have AFAIK) Didn't knew there was a DOS-only based Squeak. Any link out there to try? For GST I should note there is an interesting project using GTK which provides a gst-browser, although cannot say if now is part of GNU Smalltalk. > .- All implementations uses direct pointers, (GST?) Thank you for the pointer, really informative presentation. > - All implementations uses green threads. (VAST? MT?) Thank you, added with credits to https://github.com/hernanmd/falsehoods_smalltalk Cheers, Hernán |
Little Smalltalk Re: Falsehoods programmers believe about Smalltalk
|
|
On 22/01/2019, 02:03, "Hernán Morales Durand" <[hidden email]> wrote: Edgar @morplenauta |
SmallWorld (was Re: Little Smalltalk)
|
|
I was pleased to discover the other day "SmallWorld", Tim Budd's own
derivative of Little Smalltalk. Russell Allen has recently been keeping it running. It looks like SmallWorld was written in 2004; Russell Allen's version was updated most recently in 2015. - Tim Budd's (not completely functional) page on SmallWorld: http://web.engr.oregonstate.edu/~budd/SmallWorld/ReadMe.html - Russell Allen's SmallWorld: https://github.com/russellallen/SmallWorld I've been experimenting with using high-level languages to write tiny VMs (direct interpreter, JIT, and partial-evaluating JIT) compatible with SmallWorld images. Cheers, Tony On 1/22/19 9:09 AM, Edgar J. De Cleene wrote: > > > > On 22/01/2019, 02:03, "Hernán Morales Durand" <[hidden email]> > wrote: > > > Little Smalltalk is a good example of taking a textual > representation > and bootstrapping an image from it. > > Yes, sadly some of the LS implementations were lost in time and > need to be tracked now, _http://www.littlesmalltalk.org_ is now > chinese thing, PDST and Parla were based in LittleSmalltalk 3 > but only accessible through archive: > http://web.archive.org/web/20051025043437/_http://www.copyleft.de/Parla/Parla.html_. > There is another implementation now : > _https://github.com/0x7CFE/llst_ however didn't checked yet. > > Hernán > https://github.com/kyle-github/littlesmalltalk works wel in Mac > with class browser view running as localhost in Firefox > > A good starting point to have a WebAssembly of it, afraid lack > the skills to made it > > Edgar > @morplenauta > > > > > |
Re: Falsehoods programmers believe about Smalltalk
|
|
In reply to this post by hernanmd
Hernán Morales Durand wrote on Tue, 22 Jan 2019 02:03:01 -0300
> > [Little Smalltalk and bootstrapping an image] > > Yes, sadly some of the LS implementations were lost in time and > need to be tracked now, http://www.littlesmalltalk.org is now > chinese thing, PDST and Parla were based in LittleSmalltalk 3 but > only accessible through archive: http://web.archive.org/web/20051025043437/http://www.copyleft.de/Parla/Parla.html. > There is another implementation now : https://github.com/0x7CFE/llst > however didn't checked yet. For those who don't know about Little Smalltalk, version 1 described in the book has some important differences from Smalltalk-80. One of them is that an object is split into pieces corresponding to the superclasses so that the offset of a variable can be the same in both a class and its subclasses. This avoid having to do a lot of recompilation when a class changes. I don't remember any details about LST2, but LST3 was a lot more like Smalltalk-80 but had all classes be instances of Class like in Smalltalk-76. That was changed in LST4 to have metaclasses like Smalltalk-80. For some reason many people prefer the simpler LST3 and use that as the starting point of their forks, of which there have been many over the years. You asked about bytecodes having a fixed size. I am not sure what you mean by that. While most Smalltalk-80 bytecodes are 1 byte long, a few are "extended" and in the case of closure bytecode in Squeak they can be up to 4 bytecodes long (bytecode 143 - push closure num copied num args blocksize). The Little Smalltalk bytecodes (also the Self ones) use an operand extension bytecode like in the old Inmos Transputer. Each bytecode has a 4 bit op code and a 4 bit value for stuff that needs one argument (like push literal) or uses the 4 bit value as its op code for stuff that needs zero arguments (like return top of stack). A special extension bytecode will combine its value with the value of the next bytecode when an 8 bit argument is needed. Or two extensions allow a normal bytecode to have a 12 bit argument. While not really fixed size, this is more regular than the Smalltalk-80 scheme. > > [Selft source to bytecode compiler in the VM] > > I always wondered about how much performance is gained moving > the all the Compiler infrastructure into the VM. There should not be any performance difference if the VM compiles instead of interpreting. And the source to bytecode translation is not in the critical path anyway since only short methods are compiled at any one time while interacting with the user that won't be able to tell the difference between 3 ms and 12 ms. > Didn't knew there was a DOS-only based Squeak. Any link out there to try? > https://web.archive.org/web/20050217200230/http://www.unicavia.com:80/Squeak/Downloads.php Note that most Smalltalks that use a command line have a few helper methods defined to make things easier, but this is just Squeak without any GUI. So navigating around and doing stuff can be really awkward. I didn't test the download link to see if the zip is actually at archive.org, but I am sure I have a copy of the binary here if needed. > For GST I should note there is an interesting project using GTK which > provides a gst-browser, although cannot say if now is part of GNU > Smalltalk. There were optional GUIs for Little Smalltalk as well and a friend added a web browser based GUI to LST4. -- Jecel |
Re: Falsehoods programmers believe about Smalltalk
|
|
In reply to this post by hernanmd
On Mon, Jan 21, 2019 at 1:13 PM Hernán Morales Durand <[hidden email]> wrote:
While there might be some implementations that use a register based bytecode set I've never heard of one. I do know of a few implementations that don't use bytecode at all.
False.
Neither. In a JIT VM most time is spent executing Smalltalk code. GC overheads vary depending on workload, typically in the range 1% to 50% (typically for specially constructed benchmarks written to stress the GC).
False.
False.
False.
False. VisualWorks uses indirection pointers. Xerox Smalltalk-80 implementations used indirection. But many modern implementations use direct pointers.
False. SmalltalkMT. But still mostly true. VAST & VW have green threads and a threaded FFI.
_,,,^..^,,,_ best, Eliot |
|
|
In reply to this post by hernanmd
Thank you for your notes Richard, it's really interesting to see not so well-known projects. I would love to have more people sharing their wisdom and knocking down more myths. Cheers, Hernán El dom., 27 ene. 2019 a las 1:17, Richard O'Keefe via Glass (<[hidden email]>) escribió:
|
Re: Falsehoods programmers believe about Smalltalk
|
In reply to this post by hernanmd
Hi Hernán,
maybe you could add more references to the readme.md? Like the ones you listed here at the beginning for inspiration. That would make the repo on GitHub more self-contained and informative. Best, Marcel -- Sent from: http://forum.world.st/Squeak-Dev-f45488.html |
«
Return to Squeak - Dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |