March 2018 - Pharo-dev - lists.pharo.org

Re: [Pharo-dev] Performances of source code search on latest Pharo

by Cyril Ferlicot

On Sun, Mar 18, 2018 at 12:56 AM, Eliot Miranda <eliot.miranda(a)gmail.com> wrote: > Hi Cyril, > Hi Eliot, > > can you use the profiler (either MessageTally, or better still, > AndreasSystemProfiler) to get a better understanding? > > e.g. > > AndreasSystemProfiler spyOn: > [SystemNavigation new browseMethodsWithSourceString: > 'Method source with it' matchCase: false] > I already did it and most of the time is spent in ZnCharacterReadStream>>nextElement. This is why I said it was probably linked to the recent changes on Pharo's streams to use Zinc ones. I don't have enough knowledge on the streams and time to help more. Here is the profiler: - 132607 tallies, 132646 msec. **Tree** -------------------------------- Process: (40s) Morphic UI Process: nil -------------------------------- 99.7% {132298ms} SystemNavigation>>browseMethodsWithSourceString:matchCase: 99.7% {132239ms} SystemNavigation>>allMethodsWithSourceString:matchCase: 99.7% {132235ms} ByteString(String)>>displayProgressFrom:to:during: 99.7% {132235ms} MorphicUIManager(UIManager)>>displayProgress:from:to:during: 99.7% {132235ms} Job>>run 99.7% {132235ms} BlockClosure>>ensure: 99.7% {132221ms} Job>>run 99.7% {132219ms} CurrentJob class(DynamicVariable class)>>value:during: 99.7% {132219ms} CurrentJob(DynamicVariable)>>value:during: 99.7% {132219ms} BlockClosure>>ensure: 99.7% {132219ms} CurrentJob(DynamicVariable)>>value:during: 99.7% {132219ms} Job>>run 99.7% {132219ms} BlockClosure>>cull: 99.7% {132219ms} SystemNavigation>>allMethodsWithSourceString:matchCase: 99.7% {132219ms} SystemNavigation>>allBehaviorsDo: 94.8% {125787ms} SystemDictionary>>allClassesDo: |94.8% {125787ms} OrderedCollection>>do: | 94.8% {125783ms} SystemDictionary>>allClassesDo: | 94.8% {125774ms} SystemNavigation>>allBehaviorsDo: | 94.8% {125759ms} SystemNavigation>>allMethodsWithSourceString:matchCase: | 88.5% {117424ms} AColorSelectorMorph class(Behavior)>>selectorsDo: | |88.5% {117418ms} MethodDictionary>>keysDo: | | 88.4% {117317ms} SystemNavigation>>allMethodsWithSourceString:matchCase: | | 88.4% {117239ms} AColorSelectorMorph class(Behavior)>>sourceCodeAt: | | 88.3% {117157ms} CompiledMethod>>sourceCode | | 88.2% {117055ms} CompiledMethod>>getSourceFromFile | | 88.2% {117053ms} BlockClosure>>on:do: | | 88.2% {117051ms} CompiledMethod>>getSourceFromFile | | 88.2% {117029ms} SourceFileArray>>sourceCodeAt: | | 88.2% {117027ms} SourceFileArray>>readStreamAt:ifPresent:ifAbsent: | | 88.2% {117019ms} SourceFileArray>>readStreamAtFileIndex:atPosition:ifPresent:ifAbsent: | | 87.9% {116565ms} SourceFileArray>>sourceCodeAt: | | 87.8% {116486ms} ChunkReadStream>>next | | 87.2% {115615ms} ChunkReadStream>>basicNextChunk | | 85.0% {112685ms} SourceFile>>next | | 84.8% {112484ms} ZnCharacterReadStream(ZnEncodedReadStream)>>next | | 84.6% {112251ms} ZnCharacterReadStream>>nextElement | | 45.5% {60313ms} primitives | | 39.1% {51860ms} ZnUTF8Encoder(ZnCharacterEncoder)>>nextFromStream: | | 39.0% {51675ms} ZnUTF8Encoder>>nextCodePointFromStream: | | 38.8% {51492ms} BinaryFileStream(AbstractBinaryFileStream)>>next | | 38.1% {50479ms} ByteArray(Collection)>>ifEmpty:ifNotEmpty: | | 37.1% {49198ms} primitives | 5.9% {7822ms} ClassOrganization>>classComment | 5.9% {7822ms} ClassOrganization>>comment | 5.9% {7822ms} RemoteString>>string | 5.9% {7820ms} SourceFileArray>>readStreamAtFileIndex:atPosition:ifPresent:ifAbsent: | 5.9% {7779ms} RemoteString>>string | 5.9% {7777ms} ChunkReadStream>>next | 5.8% {7734ms} ChunkReadStream>>basicNextChunk | 5.7% {7535ms} SourceFile>>next | 5.7% {7517ms} ZnCharacterReadStream(ZnEncodedReadStream)>>next | 5.7% {7503ms} ZnCharacterReadStream>>nextElement | 3.0% {3941ms} primitives | 2.7% {3553ms} ZnUTF8Encoder(ZnCharacterEncoder)>>nextFromStream: | 2.7% {3547ms} ZnUTF8Encoder>>nextCodePointFromStream: | 2.7% {3519ms} BinaryFileStream(AbstractBinaryFileStream)>>next | 2.6% {3446ms} ByteArray(Collection)>>ifEmpty:ifNotEmpty: | 2.5% {3361ms} primitives 4.8% {6432ms} SystemDictionary>>allTraitsDo: 4.8% {6428ms} OrderedCollection>>do: 4.8% {6428ms} SystemDictionary>>allTraitsDo: 4.8% {6428ms} SystemNavigation>>allBehaviorsDo: 4.8% {6426ms} SystemNavigation>>allMethodsWithSourceString:matchCase: 4.8% {6370ms} ClyTraitChild1 classTrait(Behavior)>>selectorsDo: 4.8% {6370ms} MethodDictionary>>keysDo: 4.8% {6362ms} SystemNavigation>>allMethodsWithSourceString:matchCase: 4.8% {6356ms} ClyTraitChild1 classTrait(Behavior)>>sourceCodeAt: 4.8% {6350ms} CompiledMethod>>sourceCode 4.8% {6344ms} CompiledMethod>>getSourceFromFile 4.8% {6344ms} BlockClosure>>on:do: 4.8% {6344ms} CompiledMethod>>getSourceFromFile 4.8% {6342ms} SourceFileArray>>sourceCodeAt: 4.8% {6342ms} SourceFileArray>>readStreamAt:ifPresent:ifAbsent: 4.8% {6342ms} SourceFileArray>>readStreamAtFileIndex:atPosition:ifPresent:ifAbsent: 4.7% {6299ms} SourceFileArray>>sourceCodeAt: 4.7% {6292ms} ChunkReadStream>>next 4.7% {6238ms} ChunkReadStream>>basicNextChunk 4.6% {6078ms} SourceFile>>next 4.6% {6063ms} ZnCharacterReadStream(ZnEncodedReadStream)>>next 4.6% {6055ms} ZnCharacterReadStream>>nextElement 2.5% {3256ms} primitives 2.1% {2797ms} ZnUTF8Encoder(ZnCharacterEncoder)>>nextFromStream: 2.1% {2785ms} ZnUTF8Encoder>>nextCodePointFromStream: 2.1% {2779ms} BinaryFileStream(AbstractBinaryFileStream)>>next 2.1% {2727ms} ByteArray(Collection)>>ifEmpty:ifNotEmpty: 2.0% {2667ms} primitives **Leaves** 51.6% {68474ms} ZnCharacterReadStream>>nextElement 42.3% {56100ms} ByteArray(Collection)>>ifEmpty:ifNotEmpty: **Memory** old +0 bytes young -1,076,072 bytes used -1,076,072 bytes free +1,076,072 bytes **GCs** full 0 totalling 0ms (0.0% uptime) incr 691 totalling 859ms (1.0% uptime), avg 1.0ms tenures 0 root table 0 overflows > > -- > _,,,^..^,,,_ > best, Eliot -- Cyril Ferlicot https://ferlicot.fr

March 18, 2018

Re: [Pharo-dev] Performances of source code search on latest Pharo

by Eliot Miranda

Hi Cyril, can you use the profiler (either MessageTally, or better still, AndreasSystemProfiler) to get a better understanding? e.g. AndreasSystemProfiler spyOn: [SystemNavigation new browseMethodsWithSourceString: 'Method source with it' matchCase: false] On Sat, Mar 17, 2018 at 4:03 PM, Cyril Ferlicot <cyril.ferlicot(a)gmail.com> wrote: > Here some new precisions: > > Time to execute: SystemNavigation new browseMethodsWithSourceString: > 'Method source with it' matchCase: false > > Pharo 6.1: 2480ms > > Pharo 7: 132041ms > > The difference seems to come from SourceFileArray>>sourceCodeAt: > > Pharo 6.1: > > sourceCodeAt: sourcePointer > "Retrieves the source code for the pointer received as argument" > > ^ self > readStreamAt: sourcePointer > ifPresent: [ :stream | stream nextChunk ] > ifAbsent: [ '' ] > > Pharo 7: > > sourceCodeAt: sourcePointer > "Retrieves the source code for the pointer received as argument" > > ^ self > readStreamAt: sourcePointer > ifPresent: [ :stream | (ChunkReadStream on: stream) next ] > ifAbsent: [ '' ] > > It seems that we have a huge performance problem at some points with > the new streams. > > On Sat, Mar 17, 2018 at 7:08 PM, Cyril Ferlicot > <cyril.ferlicot(a)gmail.com> wrote: > > HI, > > > > We have a regression on source code search performances in latest Pharo. > > > > I am on Windows 10. > > > > With Pharo 6.1 it takes a few seconds to look for a string in all the > > image source code. > > > > With the latest Pharo 7 it takes minutes. > > > > To reproduce: > > - Open a playground > > - Type something > > - Select the text, right click > > - Clic on "Code search..." -> "Method source with it" > > > > https://pharo.fogbugz.com/f/cases/21612/Performances- > regression-on-source-code-search > > > > -- > > Cyril Ferlicot > > https://ferlicot.fr > > > > -- > Cyril Ferlicot > https://ferlicot.fr > > -- _,,,^..^,,,_ best, Eliot

March 17, 2018

Re: [Pharo-dev] Performances of source code search on latest Pharo

by Cyril Ferlicot

Here some new precisions: Time to execute: SystemNavigation new browseMethodsWithSourceString: 'Method source with it' matchCase: false Pharo 6.1: 2480ms Pharo 7: 132041ms The difference seems to come from SourceFileArray>>sourceCodeAt: Pharo 6.1: sourceCodeAt: sourcePointer "Retrieves the source code for the pointer received as argument" ^ self readStreamAt: sourcePointer ifPresent: [ :stream | stream nextChunk ] ifAbsent: [ '' ] Pharo 7: sourceCodeAt: sourcePointer "Retrieves the source code for the pointer received as argument" ^ self readStreamAt: sourcePointer ifPresent: [ :stream | (ChunkReadStream on: stream) next ] ifAbsent: [ '' ] It seems that we have a huge performance problem at some points with the new streams. On Sat, Mar 17, 2018 at 7:08 PM, Cyril Ferlicot <cyril.ferlicot(a)gmail.com> wrote: > HI, > > We have a regression on source code search performances in latest Pharo. > > I am on Windows 10. > > With Pharo 6.1 it takes a few seconds to look for a string in all the > image source code. > > With the latest Pharo 7 it takes minutes. > > To reproduce: > - Open a playground > - Type something > - Select the text, right click > - Clic on "Code search..." -> "Method source with it" > > https://pharo.fogbugz.com/f/cases/21612/Performances-regression-on-source-c… > > -- > Cyril Ferlicot > https://ferlicot.fr -- Cyril Ferlicot https://ferlicot.fr

March 17, 2018

Re: [Pharo-dev] Experiment: New Download page based on Pharo Launcher

by milton mamani

I have issues with Pharo Launcher. Pharo Launcher version version 1.1.1 http://files.pharo.org/pharo-launcher/tmp-1.1.1/ version 1.1 http://files.pharo.org/pharo-launcher/1.1/ You can get these links in https://github.com/pharo-project/pharo-launcher/ OS Version Mac OS - 10.13.3 Issues Issue #1, version 1.1 does not run in Mac OS - 10.13.3 * Load P.L version 1.1 * Copy the application in `Applications` folder * Try to open the program, with right-click, open * Current result the program P.L. does not run. Issue #2 Can not open/run downloaded images * Load P.L version 1.1.1 * Copy the application in `Applications` folder * Open the program, first time the program does not run * Open again the program, now the program shows a window * Try to load a pharo image, In `Templates` >> `Official distributions`>> `Pharo 6.1 stable`. * Right-click in the pharo image, then select option `Create Image` * A popup ask for a name * Introduce one name, eg: `PImage` * A loading process appears * Try to run the image with button `Launch` * Current result the image `PImage` does not run, and P.L. 1.1.1 quit on that launch. Issue #3 Pharo debugger error appear when navigating in templates * load P.L version 1.1.1 * Open the P.L * In templates area, try to load some templates * Then select one template * A debugger error window appears with the message #selectMorphList was sent to nil * Current result: Sometimes when you try to load and select some templates an error window appear Please tell me If you need more details about these issues 2018-03-17 16:42 GMT-03:00 Stephane Ducasse <stepharo.self(a)gmail.com>: > On Sat, Mar 17, 2018 at 12:59 PM, Alexandre Bergel > <alexandre.bergel(a)me.com> wrote: > > Yesterday we also experienced many problem when using Launcher on OSX :-( > > Which version of Launcher? > Which version of your OS? > What problems? > Can we reproduce it? > > Could we get pro in reporting problems? > > > > Is the Launcher still maintained? > > Alexandre > > Christophe spent two months fixing tons of issues. > So could you report problems in a nicer way? > > > > > > > > Alexandre > > > > > >> On Mar 17, 2018, at 8:57 AM, kmo <voxkmp(a)gmail.com> wrote: > >> > >> I tried the default download of Pharo Launcher on my Windows 7 (32 bit) > PC at > >> work. It installs. Runs fine - creates images and vms but when you try > to > >> run them nothing happens. It fails to start the selected image. It just > >> fails to run the vm at all as far as I can see. > >> > >> I tried it on my Windows 10 (32 bit) laptop and got exactly the same > result. > >> > >> This is not important for me as I use pharo on Linux - but I thought I'd > >> point it out. If having the launcher as the default download causes more > >> problems than the traditional vm and image download then perhaps it > should > >> be reconsidered. A pity since I love the launcher. > >> > >> > >> > >> -- > >> Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837. > html > >> > > > > > >

March 17, 2018

Re: [Pharo-dev] Adapt ImageReadWriter hierarchy to need only pure binary streams

by Stephane Ducasse

Tx sven On Sat, Mar 17, 2018 at 2:28 PM, Ben Coman <btc(a)openinworld.com> wrote: > > > On 17 March 2018 at 20:28, Sven Van Caekenberghe <sven(a)stfx.eu> wrote: >> >> https://pharo.manuscript.com/f/cases/21611 >> >> With the new File/Stream changes in Pharo 7, ImageReadWriter and >> subclasses broke because these classes are not yet capable of working with >> simple binary streams. >> >> I created a new test case (attached) to catch the functional degradation. >> For each image type (class) there are 2 tests, IO in memory and IO to file. >> I have a bunch of changes that make all tests pass. They clean up some >> strange coding as well. >> I will make a proper PR later. > > > Thanks for everyone's efforts working on this. > > cheers -ben

March 17, 2018

Re: [Pharo-dev] Experiment: New Download page based on Pharo Launcher

by Stephane Ducasse

On Sat, Mar 17, 2018 at 1:26 PM, CÃ©drick BÃ©ler <cdrick65(a)gmail.com> wrote: > Little feedback too. > > I was surprised by the new download as some students had to install on > windowsâ¦ hopefully, I had old win versions but it has disturbed me :) > Still handy and straightforward to just put an image on the VM. > > Butâ¦this being said, I just tried on Mac and this works nicely and this is > true that launcher makes the management of image clearer and easier.. > I donât know about windows but I have to care about as 98% of students > uses Pharo on windows... > > I was thinking of some adjustments: > 1) maybe a default/advanced mode on the launcher to show only, as default, > Mooc/PBE5/Official distributions. > did you see that the Launcher already includes this? There is a Pharo Mooc item. > 2) a proper icon name to display in OSâs task bars. > When several image are opened, they should have different icons, at least > different name when hovering, and not only Pharo. > The PharoLauncher should be named Pharo Launcher :). Now we should get a nice icon. > > A more general feedback is about *cloud drive OSâs* (iCloud Drive, > Dropbox, â¦). > I have 3 macâs (pro and personal). I put often files in a directory that > are synchronized across these devices. > - sometimes I get a message that an image are in different states > - sync image might be long (+changes) depending on network conditions > > I think the question is important as the launcher download in the Document > folder (OSX) and this directory is synchronized by default. > What do you think about such common situation ? What are your common > practices ? > - lets says on two Macs and works on both > - or I use a dropbox folder > > => cloud drive are cool and my impression is that we could take a better > advantage of them for data that could be shared between (personal) images > (Play-Cache ? Packages-cache ? â¦ ) > ==> would it be possible to have several files for one image (chunked over > time so only the last chunk is modified and then less file would be > synchronized) - (I donât know at all if possible) ? > ==> same for changesâ¦ but this one should be easier I think. > > => I also find longer to share pharo system the old ways - zipping (or > not) via USB - I think git integration produces a large number of files > that relies slows down system copy paste (if only image this is ok)â¦ > > > My 2 cents, > > CÃ©drick > > PS: my experiments to make students use Pharo are very coolâ¦ I need to > adjust my presentations/lecture but already I think I succeeded in getting > more attraction (and potential results) than what I tried last year with > whatever technologies (Python especially). I still use AppInventor and > Arduino.cc. > > PS2: I just like what Sean said (but might be frightening at first for > newcomers ^^) *Pharo all the way down * > > > But just in case we step into the quicksand even further, in this dystopian > > future you can just copy paste this definition: > > Pharo is a Pharo app that manages Pharo images, which can be used to make > > Pharo apps, which can be used to manage Pharo images, [which can be used to > > make Pharo apps, which can be used to manage Pharo images]* > > > Haha, yes it is so fluid it cannot be described only be experienced > through prolonged exposure, it is truly magical. > > Actually, that is not so far from the truth. > >

March 17, 2018

Re: [Pharo-dev] Experiment: New Download page based on Pharo Launcher

by Stephane Ducasse

On Sat, Mar 17, 2018 at 12:59 PM, Alexandre Bergel <alexandre.bergel(a)me.com> wrote: > Yesterday we also experienced many problem when using Launcher on OSX :-( Which version of Launcher? Which version of your OS? What problems? Can we reproduce it? Could we get pro in reporting problems? > Is the Launcher still maintained? Alexandre Christophe spent two months fixing tons of issues. So could you report problems in a nicer way? > > Alexandre > > >> On Mar 17, 2018, at 8:57 AM, kmo <voxkmp(a)gmail.com> wrote: >> >> I tried the default download of Pharo Launcher on my Windows 7 (32 bit) PC at >> work. It installs. Runs fine - creates images and vms but when you try to >> run them nothing happens. It fails to start the selected image. It just >> fails to run the vm at all as far as I can see. >> >> I tried it on my Windows 10 (32 bit) laptop and got exactly the same result. >> >> This is not important for me as I use pharo on Linux - but I thought I'd >> point it out. If having the launcher as the default download causes more >> problems than the traditional vm and image download then perhaps it should >> be reconsidered. A pity since I love the launcher. >> >> >> >> -- >> Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html >> > >

March 17, 2018

Re: [Pharo-dev] Experiment: New Download page based on Pharo Launcher

by Stephane Ducasse

On Fri, Mar 16, 2018 at 5:59 PM, Sean P. DeNigris <sean(a)clipperadams.com> wrote: > Ben Coman wrote >> it seems some people want to push PharoLauncher to *be* Pharo. >> To me this seems a poor strategyâ¦ This seems strange to me and highly >> likely to confuse newcomers. > > Yes, this was already discussed at length on the ML. > > The term Pharo already has a meaning: a vm + image containing both an OOP > language (is it Smalltalk or Smalltalk inspired? Please don't actually > answer that lol) and IDE written therein - or you could call it prototype > Dynabook software. > > If we now try to redefine the above (or really overload, because the former > definition will still coexist) to mean an app that manages images, it may > just be the final straw to make sure that no one ever will be able to give > any acceptable answer in less than 2038 hours to the question: What is > Pharo? ;) > > But just in case we step into the quicksand even further, in this dystopian > future you can just copy paste this definition: > Pharo is a Pharo app that manages Pharo images, which can be used to make > Pharo apps, which can be used to manage Pharo images, [which can be used to > make Pharo apps, which can be used to manage Pharo images]* look like smurf (Schroumpf :) > > > > ----- > Cheers, > Sean > -- > Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html >

March 17, 2018

Performances of source code search on latest Pharo

by Cyril Ferlicot

HI, We have a regression on source code search performances in latest Pharo. I am on Windows 10. With Pharo 6.1 it takes a few seconds to look for a string in all the image source code. With the latest Pharo 7 it takes minutes. To reproduce: - Open a playground - Type something - Select the text, right click - Clic on "Code search..." -> "Method source with it" https://pharo.fogbugz.com/f/cases/21612/Performances-regression-on-source-c… -- Cyril Ferlicot https://ferlicot.fr

March 17, 2018

Re: [Pharo-dev] Experiment: New Download page based on Pharo Launcher

by Peter UhnÃ¡k

> Runs fine - creates images and vms but when you try to > run them nothing happens. It fails to start the selected image. It just > fails to run the vm at all as far as I can see. I am also experiencing this... With ProcessWrapper the image launches maybe 1 out of 10 times... So I hacked for myself a solution based on Torsten's OS-Window (which uses WinAPI) ... but even then the launch success rate is below 90%... I guess Pharo really doesn't like talking to Windows... Peter On Sat, Mar 17, 2018 at 5:42 PM, Sean P. DeNigris <sean(a)clipperadams.com> wrote: > abergel wrote > > Yesterday we also experienced many problem when using Launcher on OSX :-( > > Hmm, I've been using exclusively daily for many months with minimal > glitches. Please post OS, Pharo system report, download method and any > other > relevant details. The major "gotchas" of the past were: 1) The app has to > be > in the "Applications" folder 2) the first time it was started you needed to > work around a signing issue e.g. via [right click]->Open->[Confirm dialog]. > The app will not open, but from now on you can just open in the normal way > (e.g. double clicking the .app). HTH > > > > ----- > Cheers, > Sean > -- > Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html > >

March 17, 2018