July 2019 - Pharo-users - lists.pharo.org

Question about Artifact and Especial Characters
by Juan Pablo Sandoval Alcocer July 17, 2019

July 17, 2019

Hi, We are using Artifact for automatically creating pdf invoices. It was working well so far, until we realize a problem with especial characters, for instance "ZariÄ". When some special characters are used, our script generate a white pdf as invoice :(. We tried to fix this using different char encodings for instance: UTF8 or Latin1. But then the special character is showed differently. [image: image.png] - Does anyone had this problem before? - Any idea of how may we address this? Thanks, Juan p.

1 0

ESUG19: Camp Smalltalk 24/25 August
by Marcus Denker July 17, 2019

July 17, 2019

As usual at ESUG, we will be holding an informal âCamp Smalltalkâ the weekend before the conference. The idea is to meet, hack, discuss. - location: ZWEIDENKER, Luxemburger Str. 72, 50674 KÃ¶ln (see https://zweidenker.de/kontakt/ <https://zweidenker.de/kontakt/> ) - dates & times: - Saturday, 24.8.2019 from 11:00 AM to 6:00 PM - Sunday, 25.8.2019 from 11:00 AM to 6:00 PM To have and idea who is coming, please fill out this doodle https://doodle.com/poll/ffkt7rqm4sfi62df <https://doodle.com/poll/ffkt7rqm4sfi62df> (just select when you will attend). FYI: Gamescom takes place in Cologne from 21.-24. August 2019. Cologne will be crowded and Hotels are expensive until the 24th.

1 0

(no subject)
by EVELYN CUSI LOPEZ July 15, 2019

July 15, 2019

Hi all, I'm working with artifacts, but I got an error. I have a pdf's generator which works very well, but recently I discovered that it has problems with special characters, specifically this: "Ä" since if some text has this symbol the pdf is generated in white, I tried some encodings to the text between them UTF8, Latin1, Latin2 and with these encodings the pdf manages to generate well; however the character does not look like it should look (It can be appear like this Ã or Ã¦). Does anyone know how to solve this problem? Thanks very much, Evelyn

1 1

[ANN] Pharo Newsletter June 2019
by Marcus Denker July 15, 2019

July 15, 2019

The Pharo Newsletter for June 2019 was send to subscribers. It is now available online: https://mailchi.mp/pharo/pharo-newsletter-june-2019 Find all the older newsletters in the archive: http://newsletter.pharo.org. You can subscribe to get the newsletter to get it as soon as is released. The July newsletter will be out soon (this week) Marcus

1 0

Re: [Pharo-users] [GSoC blog post] Binary Search Trees
by Richard O'Keefe July 15, 2019

July 15, 2019

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 > > >

2 1

[ANN] MessageFlowBrowser for Pharo 7 and 8
by Torsten Bergmann July 12, 2019

July 12, 2019

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.

4 3

Re: [Pharo-users] Test Completion has been added to Pharo 8 - please test!
by Cyril Ferlicot July 12, 2019

July 12, 2019

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

2 1

Re: [Pharo-users] [GSoC blog post] Binary Search Trees
by Ben Coman July 11, 2019

July 11, 2019

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

1 0

Linux init script
by SebastiÃ¡n Filippini July 10, 2019

July 10, 2019

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. This is my wrapper script: #! /bin/bash cd "$(dirname "$0")" ./pharo myapp.image start.st & PID=$! echo $PID > /var/run/myapp.pid The following process tree is created: Â ââbash,692 ./pharo myapp.image start.st Â âÂ Â ââpharo,701 --nodisplay myapp.image start.st Â âÂ Â Â Â Â Â ââ{pharo},730 And a /var/run/myapp.pid file is written with 692. Ok. The problem is that when I kill the process with pid 692 the process with pid 701 is not terminated. Any idea on how I can terminate the whole process tree? Best regards, SebastiÃ¡n Filippini

2 1

Re: [Pharo-users] [GSoC blog post] Binary Search Trees
by Richard Sargent July 10, 2019

July 10, 2019

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 > > >

1 0