December 2015 - Pharo-users

NeoJSON: parsing mapped nil fields

by Skip Lentz

Hi all, When parsing with NeoJSON, Iâm experiencing some problems. If I parse a JSON representation using NeoJSONReader>>next (i.e. not according to a schema), it parses nil fields just fine. But when I explicitly set a value schema, it wonât parse it in the case it is nil (or ânullâ in JSON). Is there a way to let it return nil in the case the parser encounters ânullâ, and otherwise serialize it as a value schema (e.g. a Point class). Taking the example from the paper (https://github.com/svenvc/docs/blob/master/neo/neo-json-paper.md <https://github.com/svenvc/docs/blob/master/neo/neo-json-paper.md>): (NeoJSONReader on: ânull' readStream) mapInstVarsFor: Point; nextAs: Point. Is there a way to adapt the above snippet so that it evaluates to ânilâ instead of resulting in an error? I ask because I am parsing an API response which sometimes contains a ânullâ field, and sometimes a JSON object for that field. Thanks.

Dec. 22, 2015

Re: [Pharo-users] Fossil and Filetree

by Hilaire

Hello, As for now, you have to use filetree, over fossil repo, but from Pharo you can only save to the repo, no way to fetch back I think. The GIT integration is more advanced with dedicated GIT package for Pharo. As a first iteration, I did some code modifications -- blindly I have to admit -- of the GIT package to decouple it from GIT, to later add fossil capability. Thierry was looking at integrating it in Jenkins job, to see if it is working or not. But if you are potential Pharo users, you should not waste your time on the CVS question but discover first the Pharo language and environment. There are other CVS options for Pharo you can use out of the box. Hilaire >> So all in all, it is pretty nice if one want to move to a private >> souce code management system with very little requirement. > > Where can I find more about using Fossil with Pharo? > > Sincerely, > Gour > -- Dr. Geo http://drgeo.eu http://google.com/+DrgeoEu

Dec. 22, 2015

A Suggested Promotion

by Richard Eng

I have a suggestion for future Smalltalk advocates... The advent of Pi Zero last month is an interesting opportunity. This is a very, very cheap Raspberry Pi. Anyone can afford it. I'm thinking of kids, in particular. What if we could get kids to delve into Squeak or Pharo (Squeak is better optimized for Raspberry Pi) via the Pi Zero? A competition could be set up in various countries, such as Canada, Britain, Australia, Japan, etc., where the best student submission for a Smalltalk-based Pi Zero application would win a small scholarship prize (say, $500). A panel of judges in each country would decide the winner. The contest would be much less ambitious than my now-defunct Canadian Smalltalk Competition <https://www.kickstarter.com/projects/1520902870/canadian-smalltalk-competit…> (CSC). Less control (no real need to prevent cheating). Easier to raise the prize money. The nice thing is that it would be a multi-national promotion (it's tough getting a national school competition like CSC going in other countries). Of course, because it's less ambitious, we can expect to reach fewer kids. Instead of thousands, probably more like hundreds. I could be totally wrong on this; I could be totally surprised! I might even come out of retirement <https://medium.com/@richardeng/today-is-the-first-anniversary-of-smalltalk-…> to offer some creative services (eg, promotional graphics, websites, etc.). ð Anyway, it was just an interesting thought I had. Don't know if it stands up to scrutiny. Cheers, Richard

Dec. 22, 2015

Re: [Pharo-users] Fossil and Filetree

by SaÅ¡a JaniÅ¡ka

On Uto, 2015-10-20 at 11:09 +0200, Hilaire wrote: Hello, potential Pharo user here who just installed 5.0 image on his Debian Sid and researching whether Smalltalk (Pharo) can be used as language for writing some multi-platform opens-source apps...tried/evaluated several languages, but, considering I abandon idea to use {C(++), Java}, the the most pragmatical option so far is to use Object Pascal Â and FPC/Lazarus(fpGUI), but someone recommended to inspect Smalltalk (I was tinkering with Digitalk looooooong ago while using OS2)... > I have been trying Fossil with Filetree, and I was rather please by > the mutuals use of these tools. So I may share a few feedbacks. > Fossil[1] itself is a software to manage source code development. It > is very simple to install and data is saved in a single sqlite > database. It comes with source code management, bug tracker, web > interface, wiki and timeline tools. so I'm delighted to see the talk about using Fossil for VCS tool with Pharo. > So all in all, it is pretty nice if one want to move to a private > souce code management system with very little requirement. For those Git-lovers, there is plan by Dr Richard Hipp (author of Fossil) to provide support which "would allow you to "clone" repos off of GitHub.Â Â Or to automatically sync your Fossil repositories on GitHub." (http://article.gmane.org/gmane.comp.version-control.fossil-scm .devel/580) Where can I find more about using Fossil with Pharo? Sincerely, Gour -- He who is regulated in his habits of eating, sleeping, recreation and work can mitigate all material pains by practicing the yoga system.

Dec. 22, 2015

Re: [Pharo-users] RemoteTask does not work on Macos

by Stephan Eggermont

On 21/12/15 21:17, David T. Lewis wrote: > My expectation is that a Unix VM (interpreter, Cog, Spur) running on > Mac with X11 display should work with the current RemoteTask, exactly > as on on Linux. But I have not personally tried it, and I am not really > sure what would be required to get it working on the standard released > Mac VMs. > > Sorry for the confusion. No problem. AioEventHandlerTestCase>>testEnableHandleAndDisable AioEventHandlerTestCase>>testHandleForFile use anOpenFile := SourceFiles at: 1. in pharo SourceFiles is a SourceFileArray that doesn't respond to #at: but uses #fileAt: In all three, UnixProcessAccessorTestCase>>testRedirectStdOutTo fails at self deny: OSProcess thisOSProcess stdOut closed. and also UnixProcessTestCase>>testCatAFile fails at self shouldnt: [p initialStdOut closed]. "Shared stdout with the VM, should be open" Ignoring #safeToForkSqueak crashes Pharo as expected when running the forking tests :( Stephan

Dec. 21, 2015

Re: [Pharo-users] Where is the documentation for the FFI ?

by Dimitris Chloupis

ah ok found it , its actually PharoLimbo , thanks Stef :) On Mon, Dec 21, 2015 at 10:28 PM stepharo <stepharo(a)free.fr> wrote: > there is X11 tutorial that was written by laurent laffond and that I > rewrote with asking stupi questions to igor. > It is on github ReadyToReview or something like that. > Stef > > > Le 19/12/15 12:06, Dimitris Chloupis a Ã©crit : > > I am considering porting my python bridge from sockets to a C DLL , > > but I cant find documentation for the Pharo FFI , where is it ? > > >

Dec. 21, 2015

Re: [Pharo-users] #storeOn: question

by stepharo

How this method is packaged? Because I do not want to have in the core like that especially if it add dependencies to MC Stef Le 16/12/15 22:47, Tudor Girba a Ã©crit : > Hi, > > I wanted to have an easy way to distinguish between a class that ships with Pharo and another class that is built on top. Unfortunately, I did not find a good enough heuristic (I used to rely on the repository of the package where the class is defined as being the Pharo one), so this was not added until we find a different solution. > > This is useful in several situations. The one I am in very often is to use extra tools that are in Moose to analyze Pharo. This was the case below where we needed to get the list of the storeOn: senders more accurately. > > Cheers, > Doru > > >> On Dec 16, 2015, at 10:41 PM, stepharo <stepharo(a)free.fr> wrote: >> >> hi doru >> >> what is isPharoClass? >> Why do you need it? >> >> Stef >> >> Le 9/12/15 21:49, Tudor Girba a Ã©crit : >>> I created an issue with a slice: >>> https://pharo.fogbugz.com/f/cases/17224/TBehavior-isPharo >>> >>> Cheers, >>> Doru >>> >>> >>>> On Dec 9, 2015, at 3:33 PM, Tudor Girba <tudor(a)tudorgirba.com> wrote: >>>> >>>> Hi again, >>>> >>>>> On Dec 9, 2015, at 3:30 PM, Tudor Girba <tudor(a)tudorgirba.com> wrote: >>>>> >>>>> Hi, >>>>> >>>>> Thanks for the question. I am at a conference, and I just took this problem to explain someone what code querying can mean :). >>>>> >>>>> Here is the result: >>>>> http://ws.stfx.eu/CCSNJVWX3JKS >>>>> >>>> I just realized that isPharoClass only exists in my image for the moment :) >>>> >>>> Doru >>>> >>>> >>>>> <storeOn.png> >>>>> >>>>> I think this is how we should document our communication because now we have the tools. >>>>> >>>>> Cheers, >>>>> Doru >>>>> >>>>> >>>>>> On Dec 3, 2015, at 7:00 AM, Werner Kassens <wkassens(a)libello.com> wrote: >>>>>> >>>>>> Hi Stephane, >>>>>> there are 89 senders of #storeOn: minus around 30 implementations of #storeOn: minus around 10 tests, also things like the PharoChangesCondenser use it. then there are 26 senders of #storeString (essentially the same functionality implemented via #storeOn:, but returns a string) eg used by the StartupPreferencesLoader. and there are several senders of #store: that are not implementations of #storeOn:. iow deprecating #storeOn: looks a bit complicated to me and is obviously way above my head. >>>>>> werner >>>>>> >>>>>> On 11/30/2015 09:42 PM, stepharo wrote: >>>>>>> It would be good to check who is using that and evaluate if we can >>>>>>> remove it. >>>>> -- >>>>> www.tudorgirba.com >>>>> >>>>> "Every thing should have the right to be different." >>>>> >>>>> >>>>> >>>> -- >>>> www.tudorgirba.com >>>> >>>> "One cannot do more than one can do." >>> -- >>> www.tudorgirba.com >>> >>> "Yesterday is a fact. >>> Tomorrow is a possibility. >>> Today is a challenge." >>> >>> >>> >>> >>> >> > -- > www.tudorgirba.com > www.feenk.com > > "It's not what we do that matters most, it's how we do it." > > >

Dec. 21, 2015

Re: [Pharo-users] [squeak-dev] encoding challenge

by Robert Withers

Thank you, this is good. I was modeling DateAndTime as UTCTime, which is a specified format for two digit years: http://www.obj-sys.com/asn1tutorial/node15.html. I noticed this other time representation while lookin... GeneralizedTime: http://www.obj-sys.com/asn1tutorial/node14.html. To meet that format with padding, I decided to convert to strings. Not pretty but there is a test! :) Unfortunately the ASN.1 DER endoding for time is specific format. I coudl just throws milliseconds in an integer slot, but I wanted to meet the spec as well for a full ASN.1 tool. I am undecided whether to use ASN.1 for header encoding in SecureSession, or custom binary specification - I use the second option at this time. Yet I am in the middle of redefining that message protocol, so now would be the time to decide. Fuel and I have a date coming up, for sure. thank you, Robert On 12/21/2015 03:34 PM, Sven Van Caekenberghe wrote: > Robert, > >> On 19 Dec 2015, at 07:09, Robert Withers <robert.w.withers(a)gmail.com> wrote: >> >> A core part of communications, whether online or to persistence, is encoding, or marshalling. Let's do an example! Thank you for your thoughtful, earnest participation, as God wills it. >> >> Here are 2 methods that encode and decode a DateAndTime object, but rather poorly. I've been going without much sleep for a bit and I can't figure it out. Here is the code in hopes that some will find this an interesting exercise in team and community building. For that intention, as a beginning. What's the best solution? >> >> Let's show them how to code when given the space in the environment. Tnks! :) >> >> In class ASN1UTCTimeType.. >> >> encodeValue: anObject withDERStream: derStream >> >> | yy mo dd hh mm ss utcDateTime | >> yy := anObject year asString copyFrom: 3 to: 4. >> mo := anObject month asString padded: #left to: 2 with: $0. >> dd := anObject dayOfMonth asString padded: #left to: 2 with: $0. >> hh := anObject hour asString padded: #left to: 2 with: $0. >> mm := anObject minute asString padded: #left to: 2 with: $0. >> ss := anObject seconds asString padded: #left to: 2 with: $0. >> utcDateTime := (yy, mo, dd, hh, mm, ss, 'Z') asByteArray. >> derStream nextPutAll: utcDateTime. >> >> >> >> decodeValueWithDERStream: derStream length: length >> >> | aUTCDateTime | >> aUTCDateTime := (derStream next: length) asByteArray asString. >> ^ (DateAndTime readFromString: ( >> ((aUTCDateTime copyFrom: 1 to: 2) asInteger > 50 ifTrue: ['19'] ifFalse: ['20']), >> (aUTCDateTime copyFrom: 1 to: 2), '-', >> (aUTCDateTime copyFrom: 3 to: 4), '-', >> (aUTCDateTime copyFrom: 5 to: 6), 'T', >> (aUTCDateTime copyFrom: 7 to: 8), ':', >> (aUTCDateTime copyFrom: 9 to: 10), ':', >> (aUTCDateTime copyFrom: 11 to: 12))) >> offset: 0 >> >> >> -- >> . .. .. ^,^ robert > Like you say, this is *not* good, because it is pretty inefficient. You want to encode binary, yet you create a string representation. But even that is done badly ;-) Any you are introducing 2YK all over again. > > Anyway, the solution is to encode the timestamp using one or more numbers, directly. From your code I assume you are only interested in second precision and the UTC/GMT/Z timezone. In that case either one rather large number (seconds since X, like unix time) or two smaller numbers (julian day number and seconds since midnight) are way more efficient. > > For inspiration, have a look at Fuel's DateAndTime>>#serializeOn: (that also encodes nanoseconds and the timezone). > > HTH, > > Sven > > > -- . .. ... ^,^ robert Go Panthers!

Dec. 21, 2015

Re: [Pharo-users] Pharo RDBMS Support Survey

by Esteban A. Maringolo

2015-11-26 0:02 GMT-03:00 Esteban A. Maringolo <emaringolo(a)gmail.com>: > Hello all, > > I just created a survey to collect > information about the users of RDBMS in Pharo. > > Please promote it in the social networks, so I can get as many > responses as possible. > I'll share the results in a month or before if I get 100 responses. :) As promised, the survey was closed with 55 responses, so we timed out and here are the results and mi interpretation of them: https://medium.com/@emaringolo/pharo-rdbms-support-survey-results-9c8f64087… Regards! Esteban A. Maringolo

Dec. 21, 2015

Re: [Pharo-users] [squeak-dev] encoding challenge

by Sven Van Caekenberghe

Robert, > On 19 Dec 2015, at 07:09, Robert Withers <robert.w.withers(a)gmail.com> wrote: > > A core part of communications, whether online or to persistence, is encoding, or marshalling. Let's do an example! Thank you for your thoughtful, earnest participation, as God wills it. > > Here are 2 methods that encode and decode a DateAndTime object, but rather poorly. I've been going without much sleep for a bit and I can't figure it out. Here is the code in hopes that some will find this an interesting exercise in team and community building. For that intention, as a beginning. What's the best solution? > > Let's show them how to code when given the space in the environment. Tnks! :) > > In class ASN1UTCTimeType.. > > encodeValue: anObject withDERStream: derStream > > | yy mo dd hh mm ss utcDateTime | > yy := anObject year asString copyFrom: 3 to: 4. > mo := anObject month asString padded: #left to: 2 with: $0. > dd := anObject dayOfMonth asString padded: #left to: 2 with: $0. > hh := anObject hour asString padded: #left to: 2 with: $0. > mm := anObject minute asString padded: #left to: 2 with: $0. > ss := anObject seconds asString padded: #left to: 2 with: $0. > utcDateTime := (yy, mo, dd, hh, mm, ss, 'Z') asByteArray. > derStream nextPutAll: utcDateTime. > > > > decodeValueWithDERStream: derStream length: length > > | aUTCDateTime | > aUTCDateTime := (derStream next: length) asByteArray asString. > ^ (DateAndTime readFromString: ( > ((aUTCDateTime copyFrom: 1 to: 2) asInteger > 50 ifTrue: ['19'] ifFalse: ['20']), > (aUTCDateTime copyFrom: 1 to: 2), '-', > (aUTCDateTime copyFrom: 3 to: 4), '-', > (aUTCDateTime copyFrom: 5 to: 6), 'T', > (aUTCDateTime copyFrom: 7 to: 8), ':', > (aUTCDateTime copyFrom: 9 to: 10), ':', > (aUTCDateTime copyFrom: 11 to: 12))) > offset: 0 > > > -- > . .. .. ^,^ robert Like you say, this is *not* good, because it is pretty inefficient. You want to encode binary, yet you create a string representation. But even that is done badly ;-) Any you are introducing 2YK all over again. Anyway, the solution is to encode the timestamp using one or more numbers, directly. From your code I assume you are only interested in second precision and the UTC/GMT/Z timezone. In that case either one rather large number (seconds since X, like unix time) or two smaller numbers (julian day number and seconds since midnight) are way more efficient. For inspiration, have a look at Fuel's DateAndTime>>#serializeOn: (that also encodes nanoseconds and the timezone). HTH, Sven

Dec. 21, 2015