- Pharo-users - lists.pharo.org

Re: [Pharo-users] Test Completion has been added to Pharo 8 - please test!

by Cyril Ferlicot

On Wed, Jul 3, 2019 at 2:29 AM Myroslava Romaniuk via Pharo-users <pharo-users(a)lists.pharo.org> wrote: > > Hi all, > > we added an intermediate version of the upcoming test completion to Pharo 8, it would be cool if you tested it and let us know the results. Here's a blog post about it - link, and here's a tweet I'd be grateful if you shared - link so perhaps people who don't read the mailing list would also see it. In the blog post you can find more info how to enable the completion in the settings and details about the whole thing in general. If something doesn't work, or if something works differently than it used to in the old completion (both if you don't like it or perhaps if you prefer it) let me know either replying here, on twitter, or submitting an issue request directly for the github repository here - link. > Hi, Since the new code completion is enable I have problem with variables completion. I saw it the most with temporary variables that are not proposed for completion anymore. This is really annoying since I often have long variables names to be explicit in my code :( I hope it can help you improve it :) > Thanks very much, > Myroslava -- Cyril Ferlicot https://ferlicot.fr

July 12, 2019

Re: [Pharo-users] [GSoC blog post] Binary Search Trees

by Richard O'Keefe

Search trees (which come in many varieties) are good for collections in which adding, deleting, and lookup are interleaved, where there is an order on the elements (if used as a set) or the keys (if used as a dictionary). So are hash tables. But (balanced) search trees have guaranteed O(lg n) worst case time, which hash tables do not. And search trees let you enumerate their contents in ascending or descending order, at any time. And search trees can support queries like - tell me all the elements {<, <=, >, >=} x - tell me all the elements between L and U - tell me the next element (before,after) x My Smalltalk library includes Sorted{Set,Bag,Dictionary} implemented as splay trees. Some time I must try an alternative. in time O(lg n + number of answers). On Thu, 11 Jul 2019 at 04:53, Alexandre Bergel via Pharo-users < pharo-users(a)lists.pharo.org> wrote: > Hello Smiljana, > > Thanks for having written down this document. I am not expert in > algorithm, so I would consider myself a simple user. I have developed > complex software for some times and I have never seen the need of having a > binary search tree. I guess this is probably partly because of my lack of > expertise in binary search tree and partly because experts in binary search > trees assume that people know what it is about and in what it is useful. > > My question is, when should a programmer ever need to use binary search > tree? Can you add some examples on what these trees are good for, and how > an average programmer should look into it. I think this will be a valuable > and easy way to expand your blog. > > Cheers, > Alexandre > > On Jul 7, 2019, at 5:12 PM, Smiljana Knezev <smilja.knezev(a)gmail.com> > wrote: > > Dear all, > > I've written about implementing binary search trees: > https://pharokeepers.github.io/pharo/2019/07/07/SmiljanaBinarySearchTreeinP… > > Feedback and opinions is always welcome :) > > Best regards, > Smiljana Knezev > > >

July 12, 2019

Re: [Pharo-users] [ANN] MessageFlowBrowser for Pharo 7 and 8

by Richard O'Keefe

Great tool. Helpful video. One thing would improve the video: remove the sound track, or replace it with spoken commentary. On Fri, 12 Jul 2019 at 04:15, Torsten Bergmann <astares(a)gmx.de> wrote: > Hi, > > I updated my MessageFlowBrowser today to work on Pharo 7 and 8, see > attached screenshot > or youtube video on https://www.youtube.com/watch?v=DRd_bzGocQg > > It now also works in Calypso and you just need to right click on a method > in the method > pane or code pane and select "Message Flow" close to Senders, Implementors > and other. > > You can load the tool using > > Metacello new > repository: 'github://astares/Pharo-MessageFlowBrowser/src'; > baseline: 'MessageFlowBrowser'; > load > > I also now moved repo from > > http://www.smalltalkhub.com/#!/~TorstenBergmann/MessageFlowBrowser > > to the new location > > https://github.com/astares/Pharo-MessageFlowBrowser > > on GitHub for future development. > > Have fun, > T. > >

July 12, 2019

Re: [Pharo-users] [ANN] MessageFlowBrowser for Pharo 7 and 8

by Ben Coman

On Fri, 12 Jul 2019 at 00:15, Torsten Bergmann <astares(a)gmx.de> wrote: > > Hi, > > I updated my MessageFlowBrowser today to work on Pharo 7 and 8, see attached screenshot > or youtube video on https://www.youtube.com/watch?v=DRd_bzGocQg > > It now also works in Calypso and you just need to right click on a method in the method > pane or code pane and select "Message Flow" close to Senders, Implementors and other. > > You can load the tool using > > Metacello new > repository: 'github://astares/Pharo-MessageFlowBrowser/src'; > baseline: 'MessageFlowBrowser'; > load > > I also now moved repo from > > http://www.smalltalkhub.com/#!/~TorstenBergmann/MessageFlowBrowser > > to the new location > > https://github.com/astares/Pharo-MessageFlowBrowser > > on GitHub for future development. Thanks Torsten. It looks really nice. cheers -ben

July 12, 2019

Re: [Pharo-users] [Pharo-dev] [ANN] OSWinSubprocess a library to spawn Windows System processes

by Mariano Martinez Peck

Hi Christophe, First of all, THANK YOU a lot for putting together this great work. This was long overdue and I never had the available time for doing it. We can discuss implementation details of OSSubprocess as much as we want, but I was personally quite satisfied with the API. So if you could come up with a similar API for Windows, I very much welcome that. And you can see even in OSSubprocess github issues that it has been asked a few times. As for the pipes...one short term solution you may want to try is to use files. If you see the whole API of OSSubprocess is polymorphic to the type of the underlying streams used for standard streams. So, for stdout, stderr, stdin etc... you can choose to either use pipes or ... regular files. In fact, in the tests, you can see 2 subclasses...one that tests all the tests with pipes and the other one with files. Finally, it explained here the differences: https://github.com/pharo-contributions/OSSubprocess#regular-files-vs-pipes Anyway, just thought I could throw my 2 cents. Congratulations, -- Mariano Martinez Peck Email: marianopeck(a)gmail.com Twitter: @MartinezPeck LinkedIn: www.linkedin.com/in/mariano-martinez-peck <https://www.linkedin.com/in/mariano-mart%C3%ADnez-peck/> Blog: https://marianopeck.wordpress.com/

July 11, 2019

[ANN] MessageFlowBrowser for Pharo 7 and 8

by Torsten Bergmann

Hi, I updated my MessageFlowBrowser today to work on Pharo 7 and 8, see attached screenshot or youtube video on https://www.youtube.com/watch?v=DRd_bzGocQg It now also works in Calypso and you just need to right click on a method in the method pane or code pane and select "Message Flow" close to Senders, Implementors and other. You can load the tool using Metacello new repository: 'github://astares/Pharo-MessageFlowBrowser/src'; baseline: 'MessageFlowBrowser'; load I also now moved repo from http://www.smalltalkhub.com/#!/~TorstenBergmann/MessageFlowBrowser to the new location https://github.com/astares/Pharo-MessageFlowBrowser on GitHub for future development. Have fun, T.

July 11, 2019

[ANN] Pharo TechTalk July 18th

by Marcus Denker

A regular chat about Pharo. Happens on Discord. Topic: How to report and fix bugs for Pharo. In a live coding session, we will together have a look at some simple fixes for Pharo8 and learn how to report bugs and do Pull Requests. - We will send an information to all subscribers of this event some hours before the talk starts. - Or just hangout in Discord, we post everything there, too https://association.pharo.org/event-3419542

July 11, 2019

Re: [Pharo-users] [GSoC blog post] Binary Search Trees

by Ben Coman

On Thu, 11 Jul 2019 at 00:53, Alexandre Bergel via Pharo-users <pharo-users(a)lists.pharo.org> wrote: > > Hello Smiljana, > > Thanks for having written down this document. I am not expert in algorithm, so I would consider myself a simple user. I have developed complex software for some times and I have never seen the need of having a binary search tree. I guess this is probably partly because of my lack of expertise in binary search tree and partly because experts in binary search trees assume that people know what it is about and in what it is useful. > > My question is, when should a programmer ever need to use binary search tree? Can you add some examples on what these trees are good for, and how an average programmer should look into it. I think this will be a valuable and easy way to expand your blog. My "computer science" is a bit rusty but Binary Search Trees and their generalisation to B-Trees is Databases 101. Almost guaranteed you are "using" them with any database you use, you're just not directly exposed to them. If you skim "Redesigning String Data Structures to Exploit Cache" (https://people.eng.unimelb.edu.au/jzobel/fulltext/acmjea10.pdf) looking at the performance graphs, for "unordered searching" the best bet for many applications is probably our hash-based Dictionary if that covers all your needs. BSTs are more memory efficient than Dictionarys, but less cache-friendly. A few applications where BSTs can be useful: * doing range searches efficiently [1] * traversing elements in order, which is more difficult with hash tables [1] [2] * good for "order" statistics [2] like: - inexact search - finding closest lower and greater elements. - determining maximum & minimum elements - determining successor & predecessor elements - range queries - easy to do with BSTs and not a natural operation with Hash Tables. * Self-Balancing BSTs guarantee operations work in O(Logn) time. Hashing has Î(1) is average time but some particular operations are costly, especially when table resizing happen [2] but also they are slower so if the data is sufficiently random simple BSTs may do. [1] https://stackoverflow.com/questions/371136/binary-trees-vs-linked-lists-vs-… [2] https://www.geeksforgeeks.org/advantages-of-bst-over-hash-table/ This side-related article... "Number crunching: Why you should never, ever, EVER use linked-list in your code again" (https://kjellkod.wordpress.com/2012/02/25/why-you-should-never-ever-ever-us…) indicates that for many In-Memory operations you are better using sorted-Arrays (e.g. SequenceableCollection>>#findBinary:) than cache-invalidating pointer-base structures like Lists & Trees - thus the strong association of BSTs & BTrees with slower access disk-based databases. Also notice that additional BST features closely match common database operations. cheers -ben

July 11, 2019

Re: [Pharo-users] Linux init script

by K K Subbu

On 10/07/19 10:30 PM, SebastiÃ¡n Filippini wrote: > Hello everyone, > > I'm trying to write a Linux init script for a Pharo application. I'm > using a bash script wrapper to launch pharo. This script executes the > pharo script provided by get.pharo.org. > > ./pharo myapp.image start.st & > PID=$! > echo $PID > /var/run/myapp.pid You could try direct exec instead of spawning a subshell. E.g. echo $$ >/var/run/myapp.id exec ./pharo myapp.image start.st HTH .. Subbu

July 10, 2019

Re: [Pharo-users] [GSoC blog post] Binary Search Trees

by Richard Sargent

On Wed, Jul 10, 2019 at 9:53 AM Alexandre Bergel via Pharo-users < pharo-users(a)lists.pharo.org> wrote: > Hello Smiljana, > > Thanks for having written down this document. I am not expert in > algorithm, so I would consider myself a simple user. I have developed > complex software for some times and I have never seen the need of having a > binary search tree. I guess this is probably partly because of my lack of > expertise in binary search tree and partly because experts in binary search > trees assume that people know what it is about and in what it is useful. > Hi Alexandre, I can tell you that GemStone/S makes extensive use of b-trees and variants in its internal data structures. Two examples are the object table itself and the index data structures. I agree that the average programmer may never have a need for them, but I don't like to close doors on (other) developers. Smiljana, Others have suggested things like the Null Object pattern for the left and right links, but an alternative or supplement to that would be a space optimizing leaf node which doesn't have instance variables for left and right sub-trees. But, don't go down the optimizing paths until you have good solutions for everything else. i.e. make right, then make it fast. I noted your code looks "C-like", which I suspect arises from being inspired by implementation in other languages. For example, I think only an empty tree should answer 0 for depth. A tree with one node should answer 1 while a tree with 2 or 3 nodes should answer 2, etc. If you drew a tree out on paper, how many levels would you see? That is the depth, in my opinion. Another example arises from one of the most powerful idioms in Smalltalk. Call it the #at:ifAbsent: pattern. (See also #detect:ifNone:) Someone else mentioned the passing around of nil and suggested you not do that. The #at:ifAbsent: pattern is the key to avoiding that: tell the receiver what *you* want to *do* if the thing can't be found by passing in your own block. Additionally, you will want your nodes to support various payloads. Right now the payload is the object held in the key. That works for identity structures, but it won't work for more advanced needs, such as one would find in a database index structure. Eventually, you will want the ability to have nodes that support additional attribute(s). I've found P. J Plauger's "0, 1, infinity" rule to be a superb design guideline. In short, once you go beyond having one of something (e.g. the key itself), you need to design things so that you can have an arbitrarily large number. The step from 1 to >1 is huge and messy if retrofitted, but easy when planned in. Eventually, you will need queries against the tree, allowing range selection, identity versus equality, etc. The tree should provide a basic infrastructure for searching, but a separate modelling of queries will be necessary. That may be out of scope for GSOC, of course. Good work and good luck with the rest of the project! > My question is, when should a programmer ever need to use binary search > tree? Can you add some examples on what these trees are good for, and how > an average programmer should look into it. I think this will be a valuable > and easy way to expand your blog. > > Cheers, > Alexandre > > On Jul 7, 2019, at 5:12 PM, Smiljana Knezev <smilja.knezev(a)gmail.com> > wrote: > > Dear all, > > I've written about implementing binary search trees: > https://pharokeepers.github.io/pharo/2019/07/07/SmiljanaBinarySearchTreeinP… > > Feedback and opinions is always welcome :) > > Best regards, > Smiljana Knezev > > >

July 10, 2019