November 2012 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] wrong FileSystem paths resolving

by Camillo Bruni

Yet another time, the truth tells me DON'T USE BASH... Or I just know too little "$*" vs "$@" did the job... On 2012-11-02, at 09:17, Pavel Krivanek <pavel.krivanek(a)gmail.com> wrote: > It works now well. Thanks. > > -- Pavel > > On Thu, Nov 1, 2012 at 11:37 PM, Camillo Bruni <camillobruni(a)gmail.com> wrote: >> I tried to fix the script AFAIK it work locally. >> can you retry the stuff on the server? >> >> best >> cami >> >> On 2012-11-01, at 22:56, Camillo Bruni <camillobruni(a)gmail.com> wrote: >>> maybe we should change vm.sh? >>> I'll have a look at it. >>> >>> >>> On 2012-11-01, at 17:15, Pavel Krivanek <pavel.krivanek(a)gmail.com> wrote: >>> >>>> Hi, >>>> >>>> in a Jenkins job configuration I have this bash line for Pharo start: >>>> >>>> ./vm.sh "images/$IMAGE_NAME.image" "$WORKSPACE/support/pharo.tank" >>>> "$WORKSPACE/support/pharo.tankst" >>>> >>>> In the console I got this error >>>> >>>> + ./vm.sh images/PharoKernelTanker-2.0.image >>>> '/builds/jenkins/workspace/Pharo Kernel 2.0 Reload by >>>> Fuel/support/pharo.tank' '/builds/jenkins/workspace/Pharo Kernel 2.0 >>>> Reload by Fuel/support/pharo.tankst' >>>> ==== Startup Error: FileDoesNotExist: >>>> file:///builds/jenkins/workspace/Pharo Kernel 2.0 Reload by >>>> Fuel/images/Fuel/support/pharo.tank >>>> >>>> As you may see, the image argument is passed by full path but it is >>>> interpreted wrongly. Tanker uses simply 'xxx' asFileReference on >>>> argument. >>>> The problem is clearly in the spaces in the path (see >>>> "Fuel/support/pharo.tank" at the end of path). On my localmachine id >>>> fails too if I use "$PWD/pharo.tank" "$PWD/pharo.tankst" as argument >>>> and the path does contain spaces. When I have this files in the image >>>> directory (even when the image path contains spaces) and I use only >>>> names of this files without full path, they are loaded well. >>>> >>>> Cheers, >>>> -- Pavel >>>> >>> >> >> >

Nov. 2, 2012

Re: [Pharo-project] wrong FileSystem paths resolving

by Pavel Krivanek

It works now well. Thanks. -- Pavel On Thu, Nov 1, 2012 at 11:37 PM, Camillo Bruni <camillobruni(a)gmail.com> wrote: > I tried to fix the script AFAIK it work locally. > can you retry the stuff on the server? > > best > cami > > On 2012-11-01, at 22:56, Camillo Bruni <camillobruni(a)gmail.com> wrote: >> maybe we should change vm.sh? >> I'll have a look at it. >> >> >> On 2012-11-01, at 17:15, Pavel Krivanek <pavel.krivanek(a)gmail.com> wrote: >> >>> Hi, >>> >>> in a Jenkins job configuration I have this bash line for Pharo start: >>> >>> ./vm.sh "images/$IMAGE_NAME.image" "$WORKSPACE/support/pharo.tank" >>> "$WORKSPACE/support/pharo.tankst" >>> >>> In the console I got this error >>> >>> + ./vm.sh images/PharoKernelTanker-2.0.image >>> '/builds/jenkins/workspace/Pharo Kernel 2.0 Reload by >>> Fuel/support/pharo.tank' '/builds/jenkins/workspace/Pharo Kernel 2.0 >>> Reload by Fuel/support/pharo.tankst' >>> ==== Startup Error: FileDoesNotExist: >>> file:///builds/jenkins/workspace/Pharo Kernel 2.0 Reload by >>> Fuel/images/Fuel/support/pharo.tank >>> >>> As you may see, the image argument is passed by full path but it is >>> interpreted wrongly. Tanker uses simply 'xxx' asFileReference on >>> argument. >>> The problem is clearly in the spaces in the path (see >>> "Fuel/support/pharo.tank" at the end of path). On my localmachine id >>> fails too if I use "$PWD/pharo.tank" "$PWD/pharo.tankst" as argument >>> and the path does contain spaces. When I have this files in the image >>> directory (even when the image path contains spaces) and I use only >>> names of this files without full path, they are loaded well. >>> >>> Cheers, >>> -- Pavel >>> >> > >

Nov. 2, 2012

Re: [Pharo-project] Unable to use Monticello in last images from Jenkins

by Camillo Bruni

On 2012-11-01, at 15:58, Guido Chari <charig(a)gmail.com> wrote: > Both tests succeed. grml, maybe squeaksource falls under the argentinian import restriction as some sort of luxury goods :P Ok so another try with zinc? ZnClient new systemPolicy; url: 'http://smalltalkhub.com/mc/estebanlm/Voyage/main'; setAcceptEncodingGzip; beOneShot; get could there be a problem with zip support in your image/vm? Can you run all the Compression tests? thanks cami > I couldn't read at home so i couldn't test it without the proxy but i'm > "almost sure" the proxy is not explaining the issue. > > > > 2012/11/1 Camillo Bruni <camillobruni(a)gmail.com> > >> could you try the following "benchmarks" >> >> >> http://dev.nuclearrooster.com/2009/11/08/checking-gzipdeflate-server-respon… >> >> with the following urls (just 2 projects): >> http://smalltalkhub.com/mc/estebanlm/Voyage/main >> http://squeaksource.com/NativeBoost/ >> >> both support gzip answers without problems. >> >> On 2012-10-31, at 23:47, Sven Van Caekenberghe <sven(a)stfx.eu> wrote: >>> Would it be possible to test without the transparant proxy ? >>> From another location, like from home ? >>> If this would fail for everyone, we would have heard about it much >> sooner. >>> >>> On 31 Oct 2012, at 22:27, Guido Chari <charig(a)gmail.com> wrote: >>> >>>> Yes, since a couple of weeks i can't. I'm on a Unix now behind a >> transparent proxy. Same Vm, on same machine, behind the same proxy, no >> problem if strip out the gzip header from the packet. >>>> >>>> I'm using a custom personalized VM. But the only thing that is >> generating the error is the gzip header option on the http packet. Is that >> option deeply related or coupled to the vm? BTW, the error is a 400 from >> the server so i don't believe so. >>>> >>>> No one else experimenting the same as me? >>>> >>>> >>>> >>>> 2012/10/31 Camillo Bruni <camillobruni(a)gmail.com> >>>> You say you cannot download from squeaksource in the latest 2.0 image? >>>> >>>> We use that regularly, for instance for the NativeBoost build, >>>> and everything works fine (win / linux / mac) >>>> >>>> That leaves me quite puzzled about the real cause of this bug... >>>> >>>> >>>> On 2012-10-31, at 21:47, Guido Chari <charig(a)gmail.com> wrote: >>>>> Hi, >>>>> >>>>> Since a couple of weeks i'm have not been able to download packages >> with >>>>> Metacello on last image versions from Jenkins. >>>>> >>>>> Today i found that if i revert changes on httpClient method from >>>>> MCHttpRepository everything works fine again. >>>>> >>>>> I haven't research deeply on why, i'm not an expert on Zinc, but the >>>>> message that is breaking everything and making my image to receive a >> 400 >>>>> error from squeaksource is setAcceptEncodingGzip that is addinng the >> gzip >>>>> header to the requests. >>>>> >>>>> As i said, i haven't going to deep with this but as i am the only one >> who >>>>> is watching this behavior perhaps is something only for squeaksource. >>>>> >>>>> If this is actually an issue and not some fault from myself, and >> someone >>>>> give me some help, i can help with the solution. >>>>> >>>>> Cheers, >>>>> Guido. >>> >>> -- >>> Sven Van Caekenberghe >>> http://stfx.eu >>> Smalltalk is the Red Pill >>> >>> >>> >>> >> >> >>

Nov. 2, 2012

Re: [Pharo-project] wrong FileSystem paths resolving

by Camillo Bruni

I tried to fix the script AFAIK it work locally. can you retry the stuff on the server? best cami On 2012-11-01, at 22:56, Camillo Bruni <camillobruni(a)gmail.com> wrote: > maybe we should change vm.sh? > I'll have a look at it. > > > On 2012-11-01, at 17:15, Pavel Krivanek <pavel.krivanek(a)gmail.com> wrote: > >> Hi, >> >> in a Jenkins job configuration I have this bash line for Pharo start: >> >> ./vm.sh "images/$IMAGE_NAME.image" "$WORKSPACE/support/pharo.tank" >> "$WORKSPACE/support/pharo.tankst" >> >> In the console I got this error >> >> + ./vm.sh images/PharoKernelTanker-2.0.image >> '/builds/jenkins/workspace/Pharo Kernel 2.0 Reload by >> Fuel/support/pharo.tank' '/builds/jenkins/workspace/Pharo Kernel 2.0 >> Reload by Fuel/support/pharo.tankst' >> ==== Startup Error: FileDoesNotExist: >> file:///builds/jenkins/workspace/Pharo Kernel 2.0 Reload by >> Fuel/images/Fuel/support/pharo.tank >> >> As you may see, the image argument is passed by full path but it is >> interpreted wrongly. Tanker uses simply 'xxx' asFileReference on >> argument. >> The problem is clearly in the spaces in the path (see >> "Fuel/support/pharo.tank" at the end of path). On my localmachine id >> fails too if I use "$PWD/pharo.tank" "$PWD/pharo.tankst" as argument >> and the path does contain spaces. When I have this files in the image >> directory (even when the image path contains spaces) and I use only >> names of this files without full path, they are loaded well. >> >> Cheers, >> -- Pavel >> >

Nov. 1, 2012

Re: [Pharo-project] wrong FileSystem paths resolving

by Camillo Bruni

maybe we should change vm.sh? I'll have a look at it. On 2012-11-01, at 17:15, Pavel Krivanek <pavel.krivanek(a)gmail.com> wrote: > Hi, > > in a Jenkins job configuration I have this bash line for Pharo start: > > ./vm.sh "images/$IMAGE_NAME.image" "$WORKSPACE/support/pharo.tank" > "$WORKSPACE/support/pharo.tankst" > > In the console I got this error > > + ./vm.sh images/PharoKernelTanker-2.0.image > '/builds/jenkins/workspace/Pharo Kernel 2.0 Reload by > Fuel/support/pharo.tank' '/builds/jenkins/workspace/Pharo Kernel 2.0 > Reload by Fuel/support/pharo.tankst' > ==== Startup Error: FileDoesNotExist: > file:///builds/jenkins/workspace/Pharo Kernel 2.0 Reload by > Fuel/images/Fuel/support/pharo.tank > > As you may see, the image argument is passed by full path but it is > interpreted wrongly. Tanker uses simply 'xxx' asFileReference on > argument. > The problem is clearly in the spaces in the path (see > "Fuel/support/pharo.tank" at the end of path). On my localmachine id > fails too if I use "$PWD/pharo.tank" "$PWD/pharo.tankst" as argument > and the path does contain spaces. When I have this files in the image > directory (even when the image path contains spaces) and I use only > names of this files without full path, they are loaded well. > > Cheers, > -- Pavel >

Nov. 1, 2012

Re: [Pharo-project] differently organized method categories in the Debugger dnu: create process

by Marcus Denker

On Nov 1, 2012, at 10:44 PM, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> wrote: > + 1 > and it will avoid that we do work in parallel. > Marcus I will close my entry. :-) I merged in into yours as a duplicate. Marcus -- Marcus Denker -- http://marcusdenker.de

Nov. 1, 2012

Re: [Pharo-project] FileSystem class>>separator?

by Camillo Bruni

On 2012-11-01, at 22:45, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> wrote: > > On Nov 1, 2012, at 10:42 PM, Camillo Bruni wrote: > >> the separator depends on the current platform... so one or the other way >> you'll have to go via an instance (repeating your statement :P). >> >> what did you have in mind? > > I was thinking aloud :) > >> besides with FileReference you rarely have to use separator, unlike with >> FileDirectory where you had to do tons of operations on path strings. > > I was fixing my web generator and I do a lot of string concatenation :) if you do filesystem stuff you don't need the separator... 'foo' asFileReference / 'bar' / 'goo' and '/foo/bar/' asFileReference will work regardless of the underlying filesystem unix/win/mac

Nov. 1, 2012

Re: [Pharo-project] FileSystem class>>separator?

by StÃ©phane Ducasse

On Nov 1, 2012, at 10:42 PM, Camillo Bruni wrote: > the separator depends on the current platform... so one or the other way > you'll have to go via an instance (repeating your statement :P). > > what did you have in mind? I was thinking aloud :) > besides with FileReference you rarely have to use separator, unlike with > FileDirectory where you had to do tons of operations on path strings. I was fixing my web generator and I do a lot of string concatenation :) Stef > > > On 2012-11-01, at 22:07, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> wrote: > >> Hi >> >> Right now if I want to get the separator I have to get the FileSystem instance >> >> FileSystem disk separator >> >> I was thinking that it would be good to be able to access separator without going to the instance. >> Does it make sense? >> Probably not since the separator depends on the store used and so separator would have to delegate anyway. >> >> Stef > >

Nov. 1, 2012

Re: [Pharo-project] differently organized method categories in the Debugger dnu: create process

by StÃ©phane Ducasse

+ 1 and it will avoid that we do work in parallel. Marcus I will close my entry. Stef On Nov 1, 2012, at 10:14 PM, Marcus Denker wrote: > Hello, > > To make this easy to review and integrate, what is the best is the following > > 1) add an issue to the issue tracker. > > http://code.google.com/p/pharo/issues/list > > I did it for this already for your report (to not forget it): > > http://code.google.com/p/pharo/issues/detail?id=6896 > > 2) to take advantage of an automatic test robot, attach a Slice to the bug report > > After having added the report, this is easy. In a fresh image: > -> do the change (e.g. load the .cs or .st) > -> open Monticello > -> Press the +Slice button > -> put the issue number 6896 into the field and press "grab" > -> select the slice and save it to the inbox > -> put status to "FixReviewNeeded" or wait until someone else does it > -> The famous monkey will load your slice and run all tests to see if you managed to break something essential. > > Integration (or rejection) will happen quite fast after this (within 24 hours often, within some days for sure). > > Marcus > > > > On Nov 1, 2012, at 9:32 PM, Paul DeBruicker <pdebruic(a)gmail.com> wrote: > >> Hi- >> >> When I create methods using the Debugger's 'Create' button I am never sure which categories are used in the class I'm adding the method to versus those in its super classes. In an effort to not have to think about that any more I've adapted the ClassDescription>>#allMethodCategoriesIntegratedThrough: to put the method categories that are used in the target class first, then those in the super classes after that. >> >> >> I think this should work in Pharo and Squeak but I've only tested it in Pharo 1.4. I've attached a changeset and also pasted the method below >> >> >> Paul >> >> >> allMethodCategoriesIntegratedThrough: mostGenericClass >> "Answer a list of all the method categories of the receiver and all its superclasses, up through mostGenericClass" >> >> | otherClassCategories thisClassCategories combinedClassCategories | >> otherClassCategories := OrderedCollection new. >> self allSuperclasses >> do: [ :aClass | >> (aClass includesBehavior: mostGenericClass) >> ifTrue: [ otherClassCategories addAll: aClass organization categories ] ]. >> otherClassCategories remove: 'no messages' asSymbol ifAbsent: [ ]. >> thisClassCategories := self organization categories asSortedCollection: [ :a :b | a asLowercase < b asLowercase ]. >> ^ Array >> streamContents: [ :stream | >> stream >> nextPutAll: thisClassCategories; >> nextPutAll: >> ((otherClassCategories asSet removeAllSuchThat: [ :each | thisClassCategories includes: each ]) >> asSortedCollection: [ :a :b | a asLowercase < b asLowercase ]) ] >> <ClassDescription-allMethodCategoriesIntegratedThrough.st> > > -- > Marcus Denker -- http://marcusdenker.de > >

Nov. 1, 2012

Re: [Pharo-project] FileSystem class>>separator?

by Camillo Bruni

the separator depends on the current platform... so one or the other way you'll have to go via an instance (repeating your statement :P). what did you have in mind? besides with FileReference you rarely have to use separator, unlike with FileDirectory where you had to do tons of operations on path strings. On 2012-11-01, at 22:07, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> wrote: > Hi > > Right now if I want to get the separator I have to get the FileSystem instance > > FileSystem disk separator > > I was thinking that it would be good to be able to access separator without going to the instance. > Does it make sense? > Probably not since the separator depends on the store used and so separator would have to delegate anyway. > > Stef

Nov. 1, 2012