July 2015 - Pharo-dev - lists.pharo.org

New Cog VMs available

by Eliot Miranda

... at http://www.mirandabanda.org/files/Cog/VM/VM.r3410 Note that there's a functional 64-bit Spur Linux StackInterpreter included with these VMs and a 64-bit image to test at http://mirandabanda.org/files/Cog/SpurImages. CogVM binaries as per VMMaker.oscog-eem.1427/r3410 General: Implement a machine code perform:[with:]* primitive. Eliminate the duplication of SimpleStackBasedCogit/StackToRegisterMappingCogit>>compileOpenPIC:numArgs: now that we have dead code elimination enabled. Change the type of the event buffer from int[8] to long[8], along with the matching event typedefs in sq.h (fixes 64-bit SPur issues). Add a fence before reading checkSignalRequests and useTideA in doSignalExternalSemaphores. ARM Include the Alien support in the Squeak ARM VMs now that it works to a fair extent in Newspeak. ARM Cogits: Use __aeabit_idivmod to implement machine code div/mod primitives on ARM. 64-bit Spur Fix at least one set of 64-bit issues caused by Slang. In particular the integerObjectOf: code resulted in (objectMemory integerObjectOf: MillisecondClockMask) evaluating to the -1 object, instead of the 16r1FFFFFFF object, which was the cause of the initially nresponsive 64-bit image on the real VM. Newspeak: Compile Newspeak self and super sends as clean sends as well. Slower in monomorphic case but net win for the larger benchmarks, presumably because it avoids I-cache flushes in the polymorphic case. Also means implementing polymorphic caches for clean sends will benefit all non-ordinary sends. DeltaBlue +8.2% Splay +7.6% ParserCombinators +4.7% Richards +0.5% SlotRead (replaced with self send) -17.6% Spill ReceiverResultReg before runtime call to fix pushEnclosingObject on ARM. (ReceiverResultReg is edx/caller-saved on IA32 but r7/callee-saved on ARM.) Fix some primitiveFail[For:] calls in potential mirror primitives that did not return after falure, and could hence continue to do damage after detecting a failure condition. Fix the 64-bit Newspeak builds' plugins.ext Integrate Ryan's code for ARM call-out and callback support in the Alien plugin. Implement the relevant implicitReceiveCacheAt: for the ARM Cogits. ThreadedFFIPlugin: Allow ffiCreateIntegralResultOop:ofAtomicType:in: to be inlined. Reduce the number of tests leading to the common case of an integral return in ffiCalloutTo:SpecOnStack:in: etc. Refactor the remapOop:in: idiom up into InterpreterPlugin and use it in the ThreadedFFIPlugin. Hence avoid unnecessary push/popRemappableOop calls. -- _,,,^..^,,,_ best, Eliot

July 19, 2015

Re: [Pharo-dev] millisecondClockValue

by Eliot Miranda

Hi Alain, that's actually a good question. The issue is whether the millisecond clock should start from zero or not. Clearly deriving it from the 64-bit microsecond clock is a win. It won't wrap (not for 50,000 years at least). But the current clock starts at 0 when the image starts up. If one were to continue to zero the millisecond clock on image startup won't overflow into large integers for 12.4 days, but it isn't really a clock. Personally my preference is to indeed implement millisecondClockValue as microsecondClockValue // 1000, and simply live with the fact that its always a large integer. On Sat, Jul 18, 2015 at 6:51 AM, Alain Plantec via Pharo-dev < pharo-dev(a)lists.pharo.org> wrote: > > > ---------- Forwarded message ---------- > From: Alain Plantec <alain.plantec(a)yahoo.com> > To: Pharo Development List <pharo-dev(a)lists.pharo.org> > Cc: > Date: Sat, 18 Jul 2015 15:50:46 +0200 > Subject: millisecondClockValue > Hello, > > Is there any reason why we do not use microsecondClockValue to compute > millisecondClockValue ? > Other that "because someone has to do it" I mean. > Iâve changed Time>>millisecondClockValue to be computed from > microsecondClockValue > and modulo change of the event stamp initialization, it seems to work > normally. > > Cheers > Alain > > > > -- _,,,^..^,,,_ best, Eliot

July 19, 2015

Re: [Pharo-dev] Edge bundles

by Ben Coman

On 18 Jul 2015 9:37 pm, "stepharo" <stepharo(a)free.fr> wrote: > > <marketing> > Think Pharo not Smalltalk. > Smalltalk is dead. Pharo is kicking. > Do not expect that people will not look at you strangely when you mention smalltalk. > And the twist is that Pharo is inspired by Smalltalk > Stef > </marketing> In general I follow this rule but I know Craig does a lot with Squeak so I was talking more broadly. Cheers -ben > Le 18/7/15 15:24, Ben Coman a Ã©crit : >> >> On Sat, Jul 18, 2015 at 7:34 PM, Craig Latta <craig(a)netjam.org> wrote: >>> >>> Hi Alexandre-- >>> >>>>> Very nice! I would love to use this for my object memory >>>>> visualizations[1], so that they can be recomputed live. Have you >>>>> done any work toward hyperbolic display, as with Walrus[2]? The >>>>> datasets I work with are sufficiently large that such a technique >>>>> is indispensable. >>>> >>>> We can discuss about this at esug. You are coming right? >>> >>> I was at the first International Conference on Live Coding[3] >>> instead this year, which happened at the same time. A great gathering! I >>> hope there's more overlap between that, ESUG, and the International >>> Workshop on Live Programming[4] in the future >> >> I would be interested to know your impressions and potential for >> Smalltalk here. What is that community's current exposure to >> Smalltalk and what are their competing technologies ? >> cheers -ben >> >> >>> [1] http://netjam.org/context/viz >>> [2] http://www.caida.org/tools/visualization/walrus >>> [3] http://iclc.livecodenetwork.org >>> [4] https://liveprogramming.github.io/2013 >> >> >> > >

July 19, 2015

Re: [Pharo-dev] Edge bundles

by Craig Latta

Hoi Stef-- > Do not expect that people will not look at you strangely when you > mention smalltalk. Actually, the people at ICLC have immense respect for Smalltalk and its current developers. It was refreshing to skip right past any sort of apologetics and get right to the doing. -C -- Craig Latta netjam.org +31 6 2757 7177 (SMS ok) + 1 415 287 3547 (no SMS)

July 18, 2015

Re: [Pharo-dev] Pharo 64bit planned?

by Eliot Miranda

On Fri, Jul 17, 2015 at 4:31 PM, Eliot Miranda <eliot.miranda(a)gmail.com> wrote: > Hi David, > > On Fri, Jul 17, 2015 at 3:22 PM, David <stormbyte(a)gmail.com> wrote: > >> I did not see any progress over the time in making a (stable) release >> of pharo in 64 bit flavor. >> >> For servers, it is a common practice (and also a good one if no 32bit >> executables are planned) to completelly disable 32bit support (by not >> compiling any library in 32bit, disabling emulation of 32bit in >> kernel), and also performance, as under 64bit, 32bit code runs slightly >> slower. >> >> The above make enough reasons to make it interesting to have a 64bit >> version of pharo, but... Is it intended/in progress? Or nothing really >> at the moment? >> > > 64-bit development of Spur is in progress. There is a somewhat functional > 64-bit Squeak Spur image and a functional Linux 64-bit Stack > interpreter. Coincidentally I'm currently working on an input event > processing bug which only occurs in the real VM. You may know that the > next release of Pharo (Pharo 6?) is intended to be a Spur release. Esteban > Lorenzano is working on the Pharo Spur bootstrap. I am working in Squeak > and my priorities are first, to get the Spur Squeak 64-bit image working > fully on the Stack Interpreter and then to work on an x64 JIT VM. I expect > that by the time Pharo 6 is ready to release, the 64-bit version will also > be ready. > and I've understood and fixed the event processing bug (which was nothing to do with 64-bits) so I can say that the Squeak Spur 64-bit Linux StackInterpreter is functional (again). Hope this helps. > -- > _,,,^..^,,,_ > best, Eliot > -- _,,,^..^,,,_ best, Eliot

July 18, 2015

Re: [Pharo-dev] Edge bundles

by Alexandre Bergel

> Please contact me privately; perhaps we can meet up somewhere in > Europe before the next ESUG. I will be at VISSOFT/ICSME end of september in Bremen, Germany Maybe SPLASH/OOPSLA in at the end of October in Pittsburg SANER beginning of March 2016 If people are around, let me know!! Cheers, Alexandre -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.bergel.eu ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.

July 18, 2015

Re: [Pharo-dev] [Slots] Slides from ESUG: Variables in Pharo5

by Marcus Denker

> On 17 Jul 2015, at 15:05, Marcus Denker <marcus.denker(a)inria.fr> wrote: > > Hi, > > The slides from the second talk are online: > > http://www.slideshare.net/MarcusDenker/variables-in-pharo5 > Friday afternoon (a bit dead by the head) Norbert and I implemented the example I mentioned in the talk (not the slides) of a slot that allows you to use as many ivars as you like but not using a dictionary (slow) but an array: https://pharo.fogbugz.com/f/cases/15995/UnlimitedInstanceVariableSlot There is a hidden base-slot that contains an array, and for every UnlimitedInstanceVariableSlot in the hierarchy it reserves an offset there. read/write are then using at: and at:put. Simple test is green, but there needs some more work to be done (see todo method) e.g. related to hierarchy changes, instance migration and code generation. Marcus

July 18, 2015

millisecondClockValue

by Alain Plantec

Hello, Is there any reason why we do not use microsecondClockValue to compute millisecondClockValue ? Other that "because someone has to do it" I mean. Iâve changed Time>>millisecondClockValue to be computed from microsecondClockValue and modulo change of the event stamp initialization, it seems to work normally. Cheers Alain

July 18, 2015

RPackage classes name

by Franck Warlouzet

Hello, Currently I am working on groups in Nautilus, and I am actually reimplementing them because I could not just fix them (There are a lot of bugs which make them almost unusable in practice), it is too complicated for nothing. Doing this I somehow have to use RPackage announcements and I am a little bit confused by their name. When you remove a package, an announcement RPackageUnregistered is raised, but I do not know if there is a difference with RPackageRemoved (which by the way does not exist, but I was expecting a name like that). What is this registration thing ? There is no RPackageRegistered but there is RPackageCreated. So I am confused. Does someone know if I should create an announcement RPackageRemoved or rename RPackageCreated into RPackageRegistered ? It does not seem consistent and so it is confusing. Or can someone explain to me the notion of registration for the RPackages ? Thanks, Franck

July 18, 2015

Re: [Pharo-dev] What happend to my smalltalkhub project?

by stepharo

Hi Nicolai Argh. Yes this is probably a smalltakhub bug. Esteban is travelling to south of italy. I hope that he will get some internet connection. Stef > I had at least one project at smalltalk hub > NicolaiHess/AthensSketch > > But now, I don't see any project, no watchlist > and the list of activities only lists the last three weeks. > > ? > > > nicolai

July 18, 2015