February 2013 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] About condensing sources, becoming CompiledMethods and VM crashes

by Guillermo Polito

'cause you don't know if the method has trailer or not I think... There are empty trailers and trailers embedding source (directly in the image)... isn't it? On Wed, Feb 20, 2013 at 10:52 AM, Marcus Denker <marcus.denker(a)inria.fr>wrote: > > On Feb 20, 2013, at 10:45 AM, Guillermo Polito <guillermopolito(a)gmail.com> > wrote: > > Hi! > > There is this bug open I was taking a look yesterday: > > > http://code.google.com/p/pharo/issues/detail?can=2&start=0&num=100&q=Milest… > > The condense sources compacts the changes and source files, and adds a new > method trailer to a compiled method with it's new source pointer. In order > to do that, it does a #becomeForward: to the method. > > CompiledMethod>>setSourcePointer: srcPointer > "We can't change the trailer of existing method, since > it could have completely different format. Therefore we need to > generate a copy with new trailer, containing an scrPointer, and then > #become it" > | trailer copy | > trailer := CompiledMethodTrailer new sourcePointer: srcPointer. > copy := self copyWithTrailerBytes: trailer. > >>> self becomeForward: copy. > ^ copy > > So far, with simple methods, so good. > > However, there are cases, in which becoming the compiled method, breaks > processes, sometimes getting a crash in the VM. For example: > > (Delay class>>#handleTimerEvent) setSourcePosition: 200 inFile: 1 > > (ProcessorScheduler class>>#idleProcess) setSourcePosition: 200 inFile: 1 > > My assumption is that the #becomeForward: of compiled methods do not get > well with the stack mapping in the vm... But just guessing. > > > I wonder why we don't just update the pointer. > That is: we keep the method the same, rewrite the pointer, and nothing > else. > > Marcus > >

Feb. 20, 2013

Re: [Pharo-project] [Jenkins][2.0] We are Green!

by Tudor Girba

Great job! Doru On Wed, Feb 20, 2013 at 8:49 AM, Marcus Denker <marcus.denker(a)inria.fr>wrote: > > on mac and linux, on windows there are just the NB related tests failing > that are known to be failing > (we should add them as known failures): > > > https://ci.inria.fr/pharo/job/Pharo-2.0-Tests/lastCompletedBuild/testReport/ > -- www.tudorgirba.com "Every thing has its own flow"

Feb. 20, 2013

Re: [Pharo-project] Weird behavior in Nautilus when doing a method rename

by Marcus Denker

On Feb 20, 2013, at 10:53 AM, "phil(a)highoctane.be" <phil(a)highoctane.be> wrote: > Yes, thanks, I figured that out. A newcomer wouldn't... and renaming > is pretty common. > Yes, and a release is only once a year.. If we want to fix everything on that level, we will *never* release anything. The release is defined as something not completely unusable of the stuff we *did do*. It is not defined as "that what I would like to have". > What will happen is that people will remove the method and recreate a new one. > Version history will then go away. > Are you sure the version history is kept when you do a refactoring? > Or is the system smart enough to find out about these things? > No. But that is *not* a show stopper for 2.0 If you want to improve the system in these directions, you should join 3.0 development early. > Phil > > 2013/2/20 Esteban Lorenzano <estebanlm(a)gmail.com>: >> hi, >> >> yes, you are right.. but that is a "nice to have", not a bug... so it will wait to 3.0 :) >> In the mean time, you can do a scoped rename: you select the packages you want, then right click and "browse scoped", then you apply you rename, and it will apply the refactor in the scoped selection, not all the image. >> >> cheers, >> Esteban >> >> On Feb 20, 2013, at 10:21 AM, "phil(a)highoctane.be" <phil(a)highoctane.be> wrote: >> >>> Well, look at the screenshot then tell me that this is what I want. >>> >>> It is definitively *not* what I want. Especially unchecking that >>> endless list of methods. >>> >>> Adding a "Uncheck all" "Check all" button to the Changes Browser list >>> would help... >>> >>> Regards >>> Phil >>> >>> 2013/2/20 Camillo Bruni <camillobruni(a)gmail.com>: >>>> there is 1 certain bug, that is that you cannot see the changes of the refactoring: >>>> >>>> https://code.google.com/p/pharo/issues/detail?id=7547 >>>> >>>> the rest I would consider a bug as well. But in the terms of refactoring it might >>>> be "valid". It does preserve behavior by renaming also all implementors. >>>> >>>> How about this reasoning: >>>> >>>> Rename method >>>> => rename all senders (since you refactor) >>>> => you have to rename all implementors as well since you renamed all send sites >>>> otherwise you'll run for sure into a DNU >>>> >>>> I think I just convinced myself :D >>>> >>>> On 2013-02-20, at 09:26, "phil(a)highoctane.be" <phil(a)highoctane.be> wrote: >>>>> I was doing a Refactoring>rename of the initialize method in Nautilus >>>>> for the ClassMethodBrowser and then the changes browser proposed me to >>>>> change all classes with initialize. WTF? >>>>> >>>>> Phil >>>>> >>>> >>>> >>> <PharoScreenshot.3.png> >> >> >

Feb. 20, 2013

Re: [Pharo-project] About condensing sources, becoming CompiledMethods and VM crashes

by Marcus Denker

On Feb 20, 2013, at 10:45 AM, Guillermo Polito <guillermopolito(a)gmail.com> wrote: > Hi! > > There is this bug open I was taking a look yesterday: > > http://code.google.com/p/pharo/issues/detail?can=2&start=0&num=100&q=Milest… > > The condense sources compacts the changes and source files, and adds a new method trailer to a compiled method with it's new source pointer. In order to do that, it does a #becomeForward: to the method. > > CompiledMethod>>setSourcePointer: srcPointer > "We can't change the trailer of existing method, since > it could have completely different format. Therefore we need to > generate a copy with new trailer, containing an scrPointer, and then > #become it" > | trailer copy | > trailer := CompiledMethodTrailer new sourcePointer: srcPointer. > copy := self copyWithTrailerBytes: trailer. > >>> self becomeForward: copy. > ^ copy > > So far, with simple methods, so good. > > However, there are cases, in which becoming the compiled method, breaks processes, sometimes getting a crash in the VM. For example: > > (Delay class>>#handleTimerEvent) setSourcePosition: 200 inFile: 1 > > (ProcessorScheduler class>>#idleProcess) setSourcePosition: 200 inFile: 1 > > My assumption is that the #becomeForward: of compiled methods do not get well with the stack mapping in the vm... But just guessing. > I wonder why we don't just update the pointer. That is: we keep the method the same, rewrite the pointer, and nothing else. Marcus

Feb. 20, 2013

Re: [Pharo-project] Weird behavior in Nautilus when doing a method rename

by phil＠highoctane.be

Yes, thanks, I figured that out. A newcomer wouldn't... and renaming is pretty common. What will happen is that people will remove the method and recreate a new one. Version history will then go away. Or is the system smart enough to find out about these things? Phil 2013/2/20 Esteban Lorenzano <estebanlm(a)gmail.com>: > hi, > > yes, you are right.. but that is a "nice to have", not a bug... so it will wait to 3.0 :) > In the mean time, you can do a scoped rename: you select the packages you want, then right click and "browse scoped", then you apply you rename, and it will apply the refactor in the scoped selection, not all the image. > > cheers, > Esteban > > On Feb 20, 2013, at 10:21 AM, "phil(a)highoctane.be" <phil(a)highoctane.be> wrote: > >> Well, look at the screenshot then tell me that this is what I want. >> >> It is definitively *not* what I want. Especially unchecking that >> endless list of methods. >> >> Adding a "Uncheck all" "Check all" button to the Changes Browser list >> would help... >> >> Regards >> Phil >> >> 2013/2/20 Camillo Bruni <camillobruni(a)gmail.com>: >>> there is 1 certain bug, that is that you cannot see the changes of the refactoring: >>> >>> https://code.google.com/p/pharo/issues/detail?id=7547 >>> >>> the rest I would consider a bug as well. But in the terms of refactoring it might >>> be "valid". It does preserve behavior by renaming also all implementors. >>> >>> How about this reasoning: >>> >>> Rename method >>> => rename all senders (since you refactor) >>> => you have to rename all implementors as well since you renamed all send sites >>> otherwise you'll run for sure into a DNU >>> >>> I think I just convinced myself :D >>> >>> On 2013-02-20, at 09:26, "phil(a)highoctane.be" <phil(a)highoctane.be> wrote: >>>> I was doing a Refactoring>rename of the initialize method in Nautilus >>>> for the ClassMethodBrowser and then the changes browser proposed me to >>>> change all classes with initialize. WTF? >>>> >>>> Phil >>>> >>> >>> >> <PharoScreenshot.3.png> > >

Feb. 20, 2013

Re: [Pharo-project] subStrings: suggestion

by Frank Shearar

On 20 February 2013 08:38, Camillo Bruni <camillobruni(a)gmail.com> wrote: > > On 2013-02-20, at 09:26, Frank Shearar <frank.shearar(a)gmail.com> wrote: > >> On 20 Feb 2013, at 0:49, Chris Cunningham <cunningham.cb(a)gmail.com> wrote: >> >>> On Tue, Feb 19, 2013 at 3:28 PM, Frank Shearar <frank.shearar(a)gmail.com> wrote: >>> <snip> >>>> It is indeed strange behavior, but I think we inherited that from squeak. >>> >>> No, Squeak has (since 2005) had still another behaviour: >>> >>> ';;;;' subStrings: ';' >>> '1;2;3' subStrings: ';' #('1' '2' '3') >>> >>> (Note the ';' rather than $;, because Squeak's version takes a String >>> of separators.) >>> >>> frank >>> >>> Yes, but it still has the odd behavior: >>> >>> ';;;;' subStrings: ';' => $( ) >>> '1;2;;3' subStrings: ';' => #('1' '2' '3') >>> >>> And the separators don't have to be strings - they can be any collection of characters. >>> >>> Chris >> >> To make my point more clearly perhaps I should have shown that in Squeak you can't use a Character argument to subStrings:, in contrast to the examples. > > neither in Pharo.. so they are 100% the same, since I can go back to > Pharo 1.0 without any version history of the method. Right. I see my mistake; I muddled the Character argument to #splitOn: with the String argument to #subStrings:. Oh well, sorry for the noise. frank

Feb. 20, 2013

Re: [Pharo-project] Weird behavior in Nautilus when doing a method rename

by Esteban Lorenzano

hi, yes, you are right.. but that is a "nice to have", not a bug... so it will wait to 3.0 :) In the mean time, you can do a scoped rename: you select the packages you want, then right click and "browse scoped", then you apply you rename, and it will apply the refactor in the scoped selection, not all the image. cheers, Esteban On Feb 20, 2013, at 10:21 AM, "phil(a)highoctane.be" <phil(a)highoctane.be> wrote: > Well, look at the screenshot then tell me that this is what I want. > > It is definitively *not* what I want. Especially unchecking that > endless list of methods. > > Adding a "Uncheck all" "Check all" button to the Changes Browser list > would help... > > Regards > Phil > > 2013/2/20 Camillo Bruni <camillobruni(a)gmail.com>: >> there is 1 certain bug, that is that you cannot see the changes of the refactoring: >> >> https://code.google.com/p/pharo/issues/detail?id=7547 >> >> the rest I would consider a bug as well. But in the terms of refactoring it might >> be "valid". It does preserve behavior by renaming also all implementors. >> >> How about this reasoning: >> >> Rename method >> => rename all senders (since you refactor) >> => you have to rename all implementors as well since you renamed all send sites >> otherwise you'll run for sure into a DNU >> >> I think I just convinced myself :D >> >> On 2013-02-20, at 09:26, "phil(a)highoctane.be" <phil(a)highoctane.be> wrote: >>> I was doing a Refactoring>rename of the initialize method in Nautilus >>> for the ClassMethodBrowser and then the changes browser proposed me to >>> change all classes with initialize. WTF? >>> >>> Phil >>> >> >> > <PharoScreenshot.3.png>

Feb. 20, 2013

[Pharo-project] About condensing sources, becoming CompiledMethods and VM crashes

by Guillermo Polito

Hi! There is this bug open I was taking a look yesterday: http://code.google.com/p/pharo/issues/detail?can=2&start=0&num=100&q=Milest… The condense sources compacts the changes and source files, and adds a new method trailer to a compiled method with it's new source pointer. In order to do that, it does a #becomeForward: to the method. CompiledMethod>>setSourcePointer: srcPointer "We can't change the trailer of existing method, since it could have completely different format. Therefore we need to generate a copy with new trailer, containing an scrPointer, and then #become it" | trailer copy | trailer := CompiledMethodTrailer new sourcePointer: srcPointer. copy := self copyWithTrailerBytes: trailer. >>> self becomeForward: copy. ^ copy So far, with simple methods, so good. However, there are cases, in which becoming the compiled method, breaks processes, sometimes getting a crash in the VM. For example: (Delay class>>#handleTimerEvent) setSourcePosition: 200 inFile: 1 (ProcessorScheduler class>>#idleProcess) setSourcePosition: 200 inFile: 1 My assumption is that the #becomeForward: of compiled methods do not get well with the stack mapping in the vm... But just guessing. Any idea? Guille

Feb. 20, 2013

Re: [Pharo-project] Weird behavior in Nautilus when doing a method rename

by phil＠highoctane.be

Well, look at the screenshot then tell me that this is what I want. It is definitively *not* what I want. Especially unchecking that endless list of methods. Adding a "Uncheck all" "Check all" button to the Changes Browser list would help... Regards Phil 2013/2/20 Camillo Bruni <camillobruni(a)gmail.com>: > there is 1 certain bug, that is that you cannot see the changes of the refactoring: > > https://code.google.com/p/pharo/issues/detail?id=7547 > > the rest I would consider a bug as well. But in the terms of refactoring it might > be "valid". It does preserve behavior by renaming also all implementors. > > How about this reasoning: > > Rename method > => rename all senders (since you refactor) > => you have to rename all implementors as well since you renamed all send sites > otherwise you'll run for sure into a DNU > > I think I just convinced myself :D > > On 2013-02-20, at 09:26, "phil(a)highoctane.be" <phil(a)highoctane.be> wrote: >> I was doing a Refactoring>rename of the initialize method in Nautilus >> for the ClassMethodBrowser and then the changes browser proposed me to >> change all classes with initialize. WTF? >> >> Phil >> > >

Feb. 20, 2013

Re: [Pharo-project] Weird behavior in Nautilus when doing a method rename

by phil＠highoctane.be

2013/2/20 Camillo Bruni <camillobruni(a)gmail.com>: > there is 1 certain bug, that is that you cannot see the changes of the refactoring: > > https://code.google.com/p/pharo/issues/detail?id=7547 > > the rest I would consider a bug as well. But in the terms of refactoring it might > be "valid". It does preserve behavior by renaming also all implementors. > > How about this reasoning: > > Rename method > => rename all senders (since you refactor) > => you have to rename all implementors as well since you renamed all send sites > otherwise you'll run for sure into a DNU > > I think I just convinced myself :D > > On 2013-02-20, at 09:26, "phil(a)highoctane.be" <phil(a)highoctane.be> wrote: >> I was doing a Refactoring>rename of the initialize method in Nautilus >> for the ClassMethodBrowser and then the changes browser proposed me to >> change all classes with initialize. WTF? >> >> Phil >> > >

Feb. 20, 2013