http://www.shlomifish.org/me/blogs/ New Topical Pages and Pages for Perl Uses on Perl-Begin http://community.livejournal.com/shlomif_tech/46426.html This summarises the changes to <a href="http://perl-begin.org/">the Perl Beginners' Site</a> since the last update. We added a <a href="http://perl-begin.org/uses/email/">a page about using Perl for E-mail processing</a>, <a href="http://perl-begin.org/uses/multitasking/">a page about Perl for multitasking and networking</a>, a some topical pages: <ul> <li> <a href="http://perl-begin.org/topics/date-and-time/">Date and Time</a>. </li> <li> <a href="http://perl-begin.org/topics/references/">References in Perl</a>. </li> <li> <a href="http://perl-begin.org/topics/regular-expressions/">Regular Expressions</a>. </li> <li> <a href="http://perl-begin.org/topics/hashes/">Hashes</a>. </li> </ul> <a href="http://perl-begin.org/IDEs-and-tools/">The IDEs and Tools page</a> now contains screenshots. Many new links have been added and many typos have been corrected. Enjoy! Perl-Begin continues to be the most comprehensive first-stop for Perl Beginners. Moreover , its content is distributed under the Creative Commons Attribution License (CC-by), and <a href="http://perl-begin.org/source/">its source code is available</a> to facilitate contributions. Shlomi Fish ( shlomif@iglu.org.il ) begin open perl perl-begin source http://community.livejournal.com/shlomif_tech/46426.html http://community.livejournal.com/shlomif_tech/46426.html Tue, 16 Mar 2010 13:38:12 GMT A Perl-based →with(...) method http://community.livejournal.com/shlomif_tech/46303.html Lately, I've been writing some <a href="http://pdl.perl.org/">PDL (Perl Data Language)</a> code and noticed that I've been doing the following: <pre> my $temp_pdl = $pdl->…->final_expr(); my $result_pdl = $temp_pdl->where($temp_pdl < 0); </pre> I've been thinking if I could somehow write it like that: <pre> my $result_pdl = $pdl->…->final_expr()->with(sub { $_->where($_ > 0)}); </pre> Which seems cleaner and more elegant. I went on IRC asking about it and someone suggested <tt>rindolf: (sub { local $_ = shift; $_ * $_ * $_})-gt;( $obj->... );</tt> (up to an episolon, but naturally this reverses the order and messes with the flow of the code). Eventually <a href="http://www.shadowcat.co.uk/blog/matt-s-trout/">Matt S. Trout (mst)</a> proposed the following solutions: <pre> sub UNIVERSAL::with { local $_ = shift; my $sub = shift; return $sub->($_,@_); } my $_with = sub { local $_ = shift; my $sub = shift; return $sub->($_,@_); }; </pre> (I had figured something like that was possible previously, but I was beating around the bush.) The $_with approach is covered in <a href="http://www.shadowcat.co.uk/blog/matt-s-trout/madness-with-methods/">one of Matt's blog posts titled "Madness with Methods"</a>. I wondered if it will work with <a href="http://search.cpan.org/dist/autobox/">autobox</a> and mst said that <q>I don't know and I hope not since the UNIVERSAL approach is evil.</q>. But it does, out-of-the-box: <pre> #!/usr/bin/perl use strict; use warnings; use autobox; sub UNIVERSAL::with { local $_ = shift; my $sub = shift; return $sub->($_,@_); } my $_with = sub { local $_ = shift; my $sub = shift; return $sub->($_,@_); }; print +(10+1)->with(sub { $_ * $_; }), "\n"; print +(10+1)->$_with(sub { $_ * $_; }), "\n"; </pre> This prints 121 twice. I now wonder what to do with this knowledge. I have some aspirations of releasing it to the cpan as a mini-module. Still, it's really cool. Finally, I should note that it seems like I'm going to lose my current Planet Perl Iron Man karma due to inadequate blogging. Oh well. Shlomi Fish ( shlomif@iglu.org.il ) perl fun http://community.livejournal.com/shlomif_tech/46303.html http://community.livejournal.com/shlomif_tech/46303.html Mon, 01 Mar 2010 17:22:55 GMT Purim Special: The Ultimate Conspiracy Theory Behind the Story of the Scroll of Esther http://shlomif.livejournal.com/62243.html It's <a href="http://en.wikipedia.org/wiki/Purim">Purim</a> today, and here is my own conspiracy theory for the events detailed in <a href="http://en.wikipedia.org/wiki/Book_of_Esther">the Scroll of Esther</a> on which the holiday is based. In my conspiracy theory, Mordechai was very interested in the politics of the Persian empire, and sought ways to influence it. He got in contact with Haman, a close childhood friend of the King, and gave him a lot of advice for how to run the empire, which Haman delivered to the king by proxy. The king was so impressed by Haman's good advice that he appointed him as prime minister (or in Hebrew "second/vice to the king."). Mordechai also had proposed his cousin, Hadasah (afterwards renamed Esther) as the wife of the king as a way to further influence the king (or it is possible that Esther delivered Mordechai's advice to Haman). However, Mordechai, Esther and Haman got tired of all the proxying and so sought a way to make Mordechai second to the king instead of Haman. So they staged the rest of what happened. Then, after the night the king could not sleep and requested that the royal records be read to him and then told Haman that Mordechai should be honoured, was indicative that he also came to understand what was going on and agreed to play along. Meir Shalev in <a href="http://bookme.co.il/Books/Item_Details.aspx?Barcode=40-10976">תנ"ך עכשיו (= "the Bible Now" in Hebrew)</a> presents Mordechai as the master mind behind the scenes of the plot, but acknowledged that there are many remaining co-incidences. On the other hand, I go a step further, and make Esther and Haman his co-conspirators, and so get rid of most of the remaining co-incidences there. Happy Purim, though it seems like it's a particularly rainy one this year. Shlomi Fish ( shlomif@iglu.org.il ) bible purim http://shlomif.livejournal.com/62243.html http://shlomif.livejournal.com/62243.html Sat, 27 Feb 2010 17:12:14 GMT Review: Alternative (to Firefox) Linux Web Browsers http://community.livejournal.com/shlomif_tech/46078.html Since I've been blogging about <a href="http://perl-begin.org/">Perl</a> a lot lately (trying to maintain and advance my <a href="http://ironman.enlightenedperl.org/">Planet Perl Iron Man</a> status), I've been neglecting to blog about other things. So here's a new entry not about Perl, not even about programming - but about web-browsers. I've been using Firefox (now at version 3.6) as my primary browser on my Mandriva Linux Cooker system for a while now, and have been mostly happy with it. Today I wanted to edit <a href="http://en.wikibooks.org/wiki/How_to_Write_a_Program">this wikibook titled "How to Write a Program"</a>. It had been written awfully (see <a href="http://en.wikibooks.org/wiki/Talk:How_to_Write_a_Program">the discussion page</a>), but I was introduced to it after it was made <a href="http://en.wikibooks.org/wiki/Optimizing_Code_for_Speed">a prerquisite of a different wikibook which I've written</a>, and decided that the dependency should be kept (assuming "How to Write a Program" would be written better). Now, I could either throw away "How to Write a Program" and start from scratch, or revamp it to submission. Right now, it seems like it would be a combination of both. In any case, after editing it, I've ran into a long-term <a href="https://qa.mandriva.com/show_bug.cgi?id=57864">bug in Mandriva, or in my Mandriva setup</a> (which is exhibited in any Firefox that is running directly on it.), which I've finally reported there (though I believe I had reported it in the Firefox bugzilla too) and so had been unable to use Firefox for it. So I've been looking for an alternative browser. I ruled out <a href="http://en.wikipedia.org/wiki/Konqueror">Konqueror</a> because KHTML has been suffering from a lot of bug report, and Konqueror is just plain annoying. (WebKit has been forked from KHTML, and it proved to be more popular.). I initially ruled out <a href="http://www.opera.com/">Opera</a> because it's not open-source. So that left me only with WebKit-based browsers. I asked for recommendations and people on <a href="irc://irc.freenode.net/#ubnutu-uk">#ubuntu-uk (great channel, BTW)</a> recommended <a href="http://en.wikipedia.org/wiki/Chromium_%28web_browser%29">Chromium</a> and <a href="http://en.wikipedia.org/wiki/Epiphany_%28web_browser%29">GNOME's Epiphany</a>, and some other more obscure browsers. I've decided to check them out. I decided to go with <a href="http://en.wikipedia.org/wiki/SRWare_Iron">SRWare Iron</a> as my Chromium browser. The first thing I noticed was that the Linux download link led to a web forum post - very unprofessional. After downloading and unpacking the archive, it refused to run - <tt>/home/shlomi/apps/iron-linux/iron: error while loading shared libraries: libbz2.so.1.0: cannot open shared object file: No such file or directory</tt>. A symbolic link to libbz2.so.1.0.0 (done by root) fixed that problem, and then it crashed with an X error. However, it started the second time and seemed to run fine (don't know what the problem was). I noticed Chromium was indeed very fast, but then I ran into a few glitches: <ul> <li> No menu bar. WTF? All desktop applications, including web browsers had menu bars for generations. And I'm used to invoke the File menu and select "New Window" and "New Tab". But Chromium does not have any menus, and it appears to be by intentional design (but probably a crappy one). </li> <li> If this wasn't enough, there was another more minor but still irritating annoyance: after clicking on the URL bar and pressing double click, only the current word is selected as opposed to the Firefox behaviour of selecting the entire URL. </li> </ul> So I decided to look at Epiphany. Slower than Chromium, but maybe it will work. Or maybe it won't. The first thing I noticed was that it looked ugly - very unaesthetic and several widgets in the main window were slightly out-of-place. Its URL bar also behaved unlike Firefox in the double click respect, which was also annoying. So what can I do? I eventually decided to use Opera for editing Wikimedia wikis until the Firefox-influencing bug is resolved. It is fast, it has a menu bar, and double click in the URL bar selects the entire URL. Very nice. It's not open-source, so I'm trying not to get myself used to it, but seems like I'll need to use it. If you haven't done so already, I suggest you read <a href="http://www.joelonsoftware.com/navLinks/fog0000000262.html">the "Joel on Software" "must-read" books recommendations page</a>. I haven't read all the books there, but the page itself is also good. Quoting from it from our context: <blockquote> A few months ago when we released CityDesk, I got an email from a customer complaining that he was used to doing Alt+F, Alt+S to save files. Unfortunately due to a tiny, unnoticed bug, that keyboard shortcut saved the file and then closed it, irritatingly. I had never noticed because I'm in the habit of doing Alt+F,S to save files, not Alt+F,Alt+S -- a tiny difference -- and Alt+F,S worked fine. Once you get into the habit of doing Alt+F,Alt+S to save, it becomes so automatic you don't think of it as Alt+F,Alt+S. You think of it as save. And when you push the "save" button in your brain and the file you were working on goes away, it makes you feel like you're not in control of your environment. It's a small thing, but about the fourth time that it happens, you're going to be seriously unhappy. That's why I spent several hours tracking down this bug and fixing it. In a bizarre application of Murphy's Law, this fix led to a cascade of events that caused us to waste something like a week, but that's neither here nor there. It was worth the time spent. This is what it means to be concerned about usability. If you still think that something as small as how long you hold down the Alt key when you activate a menu command doesn't matter, well, your software is going to make people unhappy. These tiny inconsistencies are what makes Swing applications so unbearably annoying to use, and in my opinion it's why there are virtually no commercially successful Java GUI applications. </blockquote> If I can't go to "File → New Tab" in the menubar to start a new tab then I'm going to be unhappy. If I have to triple click the URL bar in order to select the entire URL, then I'm going to be unhappy. These tiny things are why I can't use either Chromium or Epiphany. (And frankly, a lack of a menubar is a huge UI screw-up, not a tiny one). At the moment, I'm just ranting, but I am planning on reporting these bugs to the bug trackers of the appropriate projects. I just hope Google won't think the lack of menu-bar in Chromium is a feature. "Welcome to Google UIs". (Welcome to Google Hell…). Shlomi Fish ( shlomif@iglu.org.il ) web browser firefox open source http://community.livejournal.com/shlomif_tech/46078.html http://community.livejournal.com/shlomif_tech/46078.html Thu, 25 Feb 2010 21:21:42 GMT Etymology of the Word "Cat" http://shlomif.livejournal.com/61969.html According to <a href="http://en.wiktionary.org/wiki/cat#Etymology_1">the Wiktionary entry of the English word "Cat"</a> (for domestic felines, etc.) it is originally derived from the Late Egyptian "čaute", feminine of "čaus 'jungle cat". I guess the Hebrew word for Cat, <a href="http://en.wiktionary.org/wiki/%D7%97%D7%AA%D7%95%D7%9C">"Chatul"</a> has a similar origin, and I never noticed the similarity between the two words until I've read the etymology in the Wiktionary. There are similar words in other European languages: <a href="http://en.wiktionary.org/wiki/chat#French">the French chat</a>, and <a href="http://en.wiktionary.org/wiki/Katze#German">the German "Katze"</a>, and the Arabic word for Cat is <a href="http://en.wiktionary.org/wiki/%D9%82%D8%B7">Qitt</a>. Make sure you avoid a <a href="http://www.shlomifish.org/humour/bits/Mastering-Cat/">useless use of "cat"</a>… Shlomi Fish ( shlomif@iglu.org.il ) linguistics cats etymology http://shlomif.livejournal.com/61969.html http://shlomif.livejournal.com/61969.html Tue, 23 Feb 2010 14:59:33 GMT New Perl Quiz-of-the-Whatever: Short Freecell Solutions http://community.livejournal.com/shlomif_tech/45610.html I published a new <a href="http://article.gmane.org/gmane.comp.lang.perl.qotw.discuss/2655">Perl Quiz-of-the-Whatever about finding short Freecell solutions</a>. Hope you enjoy thinking about it. Here is the <a href="http://perl.plover.com/qotw/">Perl Quiz-of-the-Whatever (formerly "Quiz of the Week") information.</a>. Shlomi Fish ( shlomif@iglu.org.il ) perl fc-solve freecell http://community.livejournal.com/shlomif_tech/45610.html http://community.livejournal.com/shlomif_tech/45610.html Mon, 22 Feb 2010 21:57:30 GMT Solving gaal's Circuit Riddle Using Perl http://community.livejournal.com/shlomif_tech/45409.html <a href="http://gaal.livejournal.com/234795.html">gaal (= Gaal Yahas) mentions a riddle</a> on his blog: <blockquote> There’s a box with three input signals: A, B, C. It has three output signals: ~A, ~B, ~C (that is, the negation of each input). Design a circuit satisfying the above description. You have two NOT gates, and as many AND and OR gates as you like. (If you solve this, don’t tell me the answer — I haven’t solved it yet; I just want to share the misery.) </blockquote> I decided that I will try to solve this riddle on my own and then attempt to write a solution using a Perl 5 program, which I've also thought about how to write. As it turned out, I ended up exploring and eventually completely solving the riddle using a series of Perl programs. I won't mention the final solution here, but I'll tell which programs I wrote to solve them and how I wrote them. The code can be found in <a href="http://bitbucket.org/shlomif/riddle-not-a-not-b-not-c/src/">its Mercurial repository</a> under the MIT/X11 licence, as I mentioned in <a href="http://gaal.livejournal.com/234795.html?thread=936235#t936235">a comment</a> to gaal's post. Well, first of all I decided to write a program that will take two negated expressions of AND and OR gates and see if one can arrive to the final solution using them and the given inputs (A, B and C). To do this I represented every such possible expression as its components of (A + ~A) × (B + ~B) × (C + ~C) - 8 in total which fit in an 8-bit datum, and which I indexed using an array of 256 items to see which ones exist in the population. After doing the coding I realised none of the expressions I tried worked. Then I decided to see if I there are any two possible expressions that when added to A, B an C will yield all the necessary outputs. I ran it and indeed found a single pair (and only a single pair) of such expressions (given as numbers) which I was able to analyse. I should note that the programs as I've written it using some functions and closures was too slow, so I ended up writing another Perl script to preprocess the code, and inline the functions, which has made everything much faster. (You can find this preprocessor in the repository). Then I wanted to see if I can indeed construct these expressions from A , B, and C using two negations. When I had thought about this, I believed I needed to maintain a collection of expressions' populations, and to keep how many negations were in each. But when I got to code it I thought of a simpler way: marry all the inputs using AND and OR until no more can be married, then go over the existing ones and negate each one in turn, then add the negated vector to the population and marry it with all the rest, and then do it again for all the existing inputs. Using this I was able to find a single solution. Then I had to output the solution. For that, I wrote a recursive expression dumper, which ended up outputting with many parentheses, so I had to post- process it a bit. The solution I got to was very complicated and I believe it would have taken me a long time to reach without resorting to programming. But at least it saved me a lot of time and head-scratching. Shlomi Fish ( shlomif@iglu.org.il ) perl fun http://community.livejournal.com/shlomif_tech/45409.html http://community.livejournal.com/shlomif_tech/45409.html Fri, 19 Feb 2010 13:28:13 GMT "Escape from GNU Autohell", List of Editors and IDEs and Factoids' Fortunes Collection http://community.livejournal.com/shlomif_hsite/14312.html There's <a href="http://www.shlomifish.org/humour.html#english-spelling">a new joke in the aphorisms' collection</a>: <blockquote> English spelling aims to be consistent. Publicly and methodically. </blockquote> I added <a href="http://www.shlomifish.org/humour/fortunes/shlomif-factoids.html">a new fortune cookies' collection with factoids about Chuck Norris/etc.</a>, concentrating the ones from <a href="http://www.shlomifish.org/humour/bits/facts/">the collections of facts section</a>. There are also some new fortune cookies there: <blockquote> <ul> <li>Shlomif:BTW, have you read my stories yet?</li> <li>Sjors:I haven't</li> <li>Shlomif:Ah.</li> <li>Shlomif:"If you read my stories, I'll give you 1,000,000 virtual dollars."</li> <li>Sjors:Causing me to have a lot of extra virtual time!</li> <li>Shlomif:And be virtually rich.</li> <li>Shlomif:And then you can virtually bribe virtual politicians.</li> <li>Shlomif:And buy a lot of virtual goods.</li> <li>Shlomif:LOL.</li> <li>Sjors:Then, I'd be virtually happy</li> <li>Sjors:Too bad... :P</li> <li>Shlomif:It's a virtual win-win situation.</li> <li>Shlomif:You can hire many virtual programmers to write a lot of virtual code for KMess.</li> <li>Shlomif:"My old virtual dad used to say to me: 'virtual money does not bring you virtual happiness, my virtual son.'"</li> </ul> </blockquote> There's <a href="http://www.shlomifish.org/open-source/resources/editors-and-IDEs/">a new page with a list of text editors and IDEs</a>. I also added <a href="http://www.shlomifish.org/open-source/anti/autohell/">a page titled "Escape from GNU Autohell!"</a> explaining why the GNU Autotools suck so much and why you should switch to CMake or a different better alternative. I added <a href="http://www.shlomifish.org/open-source/bits-and-bobs/greasemonkey/grease.html#self-links-for-headers">a new Greasemonkey script</a> for providing self-links for headers with an <tt>id=</tt> attribute. <a href="http://www.shlomifish.org/lecture/Freecell-Solver/project-intro/">The Freecell Solver project intro</a> now contains <a href="http://www.shlomifish.org/lecture/Freecell-Solver/project-intro/Summary.txt">a summary</a>. Shlomi Fish ( shlomif@iglu.org.il ) facts autotools www site humour fish autohell norris web homepage jokes chuck shlomi ides cmake factoids text editors http://community.livejournal.com/shlomif_hsite/14312.html http://community.livejournal.com/shlomif_hsite/14312.html Thu, 18 Feb 2010 15:40:22 GMT Article Recommendation: "What Every CSer Should Know about Floating-Point Arithmetic" http://community.livejournal.com/shlomif_tech/45104.html I've read the paper <a href="http://docs.sun.com/source/806-3568/ncg_goldberg.html">"What Every Computer Scientist Should Know About Floating-Point Arithmetic"</a> and can recommend it as it has many interesting insights. I had received quite a lot of reports for bugs in <a href="http://search.cpan.org/dist/Statistics-Descriptive/">the "Statistics-Descriptive" CPAN distribution</a> that were caused by expected floating-point behaviour. Some of them I was able to resolve using some trickery, and some others possibly can be somewhat mitigated using the insights I've learnt in the article. Shlomi Fish ( shlomif@iglu.org.il ) perl recommendation article http://community.livejournal.com/shlomif_tech/45104.html http://community.livejournal.com/shlomif_tech/45104.html Thu, 11 Feb 2010 05:24:18 GMT NYTProf-3 is Out! http://community.livejournal.com/shlomif_tech/44836.html <a href="http://blog.timbunce.org/2009/12/24/nytprof-v3-worth-the-wait/">Tim Bunce writes on his blog about the new features in Devel-NYTProf version 3</a>. Devel-NYTProf is a profiler for the <a href="http://perl-begin.org/">Perl programming language</a>, which has put all the previous attempts in profiling in the dust, and now it's even better than before. Enjoy! (Thanks to <a href="http://mail.pm.org/pipermail/sanfrancisco-pm/2010-February/002704.html">Fred Moyer's post</a> on the San-Fransisco Perl Mongers mailing list). Shlomi Fish ( shlomif@iglu.org.il ) 3 version perl profiler release nytprof http://community.livejournal.com/shlomif_tech/44836.html http://community.livejournal.com/shlomif_tech/44836.html Fri, 05 Feb 2010 16:31:51 GMT New Page about Text Editors and IDEs for Programmers http://community.livejournal.com/shlomif_tech/44605.html After answering the questions "Can anyone recommend a good text editor?" and "What is a good IDE?" a few times in the past, I set up <a href="http://www.shlomifish.org/open-source/resources/editors-and-IDEs/">a page listing some prominent text editors and Integrated Development Environments (IDEs)</a> for development on my homepage. Currently it includes only cross-platform open-source text editors and cross-platform open-source IDEs, but I'm planning to expand it as time and interest permits. Additions would be welcome. <a href="http://www.vim.org/">Vim</a> is there and so are XEmacs and <a href="http://www.eclipse.org/">Eclipse</a>, and also <a href="http://padre.perlide.org/">Padre, the Perl IDE</a> and more specialised IDEs such as <a href="http://eric-ide.python-projects.org/">the "Eric" Python IDE</a>. Shlomi Fish ( shlomif@iglu.org.il ) http://community.livejournal.com/shlomif_tech/44605.html http://community.livejournal.com/shlomif_tech/44605.html Tue, 02 Feb 2010 19:59:14 GMT פגישת קוד פתוח בתל-אביב: 14 בפברואר - vtiger CRM http://community.livejournal.com/shlomif_tech/44385.html <div dir="rtl" align="right"> <a href="http://tel.foss.org.il/">מועדון הקוד הפתוח התל-אביבי (תלוקס)</a> ייפגש שוב כדי לשמוע את <a href="http://wiki.osdc.org.il/index.php/Tel_Aviv_Meeting_on_14_February_2010">הרצאתו של רמי הדדי אודות "מערכת ניהול קשרי הלקוחות vtigerCRM</a>. ההרצאה תתקיים ביום ראשון, 14 בפברואר 2010, בשעה 18:00, באולם הולצבלט, מס' 007 במסדרון הבניינים למדעים מדויקים (שימו לב לשינוי בשעה ובמיקום משנה שעברה) באוניברסיטת תל אביב. פרטים נוספים, מפות להגעה וכיוצא בזה, ניתן למצוא <a href="http://tel.foss.org.il/">באתר</a> <a href="http://wiki.osdc.org.il/index.php/Tel_Aviv_Meeting_on_14_February_2010">ובוויקי</a>. הנוכחות בהרצאה היא חינמית ולא נדרשת הרשמה מראש. <a href="http://www.vtiger.com/">vtiger CRM</a> היא מערכת ניהול קשרי לקוחות אשר פועלת בממשק Web דרך הדפדפן. המערכת מכילה את כל הכלים הנדרשים לניהול מלא ומקיף של עסק: החל מעסק קטן של אדם אחד וכלה בעשרות עובדים. במערכת ניתן למצוא ביטוי ללקוחות, יומנים, הזדמנויות עסקיות, מכירות, שירות ועוד. המערכת מבוססת על קוד-פתוח וניתן גם לבצע בה התאמות בקלות רבה. המערכת תורגמה לשפות רבות ובכללן עברית. רמי הדדי הינו מנכ"ל אקטיבטק אשר מתמחה ביעוץ, פיתוח ואירוח של vtiger CRM. אנו תמיד מחפשים מרצים שיתנדבו לתת הרצאות בנושאים שונים הקשורים לקוד-פתוח ולמחשבים. במידה שאתם מעוניינים לתת הרצאה, או שיש לכם הצעה להרצאה שמעניינת אתכם, נשמח לשמוע ממכם. </div> Shlomi Fish ( shlomif@iglu.org.il ) http://community.livejournal.com/shlomif_tech/44385.html http://community.livejournal.com/shlomif_tech/44385.html Tue, 02 Feb 2010 18:47:42 GMT Escape from GNU Autohell! http://community.livejournal.com/shlomif_tech/44103.html I've set up <a href="http://www.shlomifish.org/open-source/anti/autohell/">a page on my web-site titled "Escape from GNU Autohell!"</a> about the evils of the GNU Autotools (GNU Autoconf/Automake/Libtool and other GNU auto-brain-damages) and why you should convert to a good alternative such as <a href="http://www.cmake.org/">CMake</a>. The page contains a list of the many disadvantages of the GNU Autotools, some refutation of common arguments against CMake, and some links. And lo and behold, some time after writing the page I spent hours dealing with yet another Autohell problem: <a href="https://qa.mandriva.com/show_bug.cgi?id=57319">"Mandriva's libtool causes "make install" of xine-lib-1.1-hg to fail"</a>. I actually had to build the Mandriva libtool several times to find the offending downstream patch, and libtool's "./bootstrap" script was so slow that I could finish running "make" and "make install" in xine-lib while waiting for it to finish. And it says a lot because xine uses GNU Autohell which slows down the build process considerably. Eventually, I found the offending patch after several hours, and I wouldn't have finished on time if I didn't cancel the "make check" test in the .rpm. Autohell must die! Shlomi Fish ( shlomif@iglu.org.il ) rants cmake build autohell gnu escape rant make autotools http://community.livejournal.com/shlomif_tech/44103.html http://community.livejournal.com/shlomif_tech/44103.html Sun, 31 Jan 2010 12:45:57 GMT Project Euler Problem #10 in Haskell, Perl and C http://community.livejournal.com/shlomif_tech/44006.html zerothorder told <a href="http://zerothorder.blogspot.com/2009/11/after-drinking-functional-immutable-and.html">how he found a solution for Project Euler's Problem 10 in Haskell</a>. The problem is "Find the sum of all primes less than 2,000,000". It is given here below: <pre> primes :: [Integer] primes = 2 : filter isPrime [3, 5 ..] where -- only check divisibility of the numbers less than the square root of n isPrime n = all (not . divides n) $ takeWhile (\p -> p*p <= n) primes divides n p = n `mod` p == 0 result = sum $ takeWhile (< 2000000) primes main = do putStrLn( show result ) </pre> He says "If this doesn't give you a nerdgasm, I don't know what will.". The problem is that this nerdgasm will last a long time. Benchmarking this program gives that 20 iterations of it run at 310 seconds - less than - about 15 seconds each (on my Pentium 4 2.4GHz machine running Mandriva Linux Cooker). So next I tried a better Haskell implmenetation that I recalled from a thread I started in the Haskell Café mailing list about implementing <a href="http://en.wikipedia.org/wiki/Sieve_of_Eratosthenes">a sieve of Eratosthenes in Haskell</a>: <pre> import Data.Int primes :: Int64 -> [Int64] primes how_much = sieve [2..how_much] where sieve (p:x) = p : (if p <= mybound then sieve (remove (p*p) x) else x) where remove what (a:as) | what > how_much = (a:as) | a < what = a:(remove what as) | a == what = (remove (what+step) as) | a > what = a:(remove (what+step) as) remove what [] = [] step = (if (p == 2) then p else (2*p)) sieve [] = [] mybound = ceiling(sqrt(fromIntegral how_much)) --main = print (length (primes 1000000)) main = print (sum (primes 2000000)) </pre> This does not involve costly operations such as modulo or division and 20 iterations of it run at 135 wallclocks seconds - over two times faster than zeroth's Haskell version. Now how about Perl? Since Perl has assignment, we have the advantage that we can create a vector of bits that we will mark with the primes by iterating over all numbers up to the root of the limit. Here is the code: <pre> #!/usr/bin/perl use strict; use warnings; use Math::BigInt lib => 'GMP'; my $limit = 2_000_000; my $primes_bitmask = ""; my $loop_to = int(sqrt($limit)); my $sum = 0; my $total_sum = Math::BigInt->new('0'); for my $p (2 .. $loop_to) { if (vec($primes_bitmask, $p, 1) == 0) { $sum += $p; my $i = $p * $p; while ($i < $limit) { vec($primes_bitmask, $i, 1) = 1; } continue { $i += $p; } } } for my $p ($loop_to .. $limit) { if (vec($primes_bitmask, $p, 1) == 0) { if (($sum += $p) > (1 << 30)) { $total_sum += $sum; $sum = 0; } } } $total_sum += $sum; print "$total_sum\n"; </pre> 20 runs of it run at 95 walclock seconds - even faster than the Haskell version. But it gets better. Since all the primes we encounter greater than 2 are not even, we can create a map of their pseduo-halves and conserve on memory and iterations. This is the Perl version: <pre> #!/usr/bin/perl use strict; use warnings; use Math::BigInt lib => 'GMP'; my $limit = 2_000_000; my $primes_bitmask = ""; my $loop_to = (int(sqrt($limit)))>>1; my $half_limit = ($limit-1)>>1; my $sum = 0+2; my $total_sum = Math::BigInt->new('0'); for my $half (1 .. $loop_to) { if (vec($primes_bitmask, $half, 1) == 0) { my $p = (($half<<1)+1); $sum += $p; my $i = ($p * $p)>>1; while ($i < $limit) { vec($primes_bitmask, $i, 1) = 1; } continue { $i += $p; } } } for my $half ($loop_to .. $half_limit) { if (vec($primes_bitmask, $half, 1) == 0) { if (($sum += (($half<<1)+1)) > (1 << 30)) { $total_sum += $sum; $sum = 0; } } } $total_sum += $sum; print "$total_sum\n"; </pre> Running this 20 times takes 73 wallclock seconds, close to half that of my Haskell version. Then I wondered how long C will take. Here is a C implementation without the halving: <pre> #include <string.h> #include <math.h> #include <stdint.h> #include <stdio.h> #define limit 2000000 int8_t bitmask[(limit+1)/8]; int main(int argc, char * argv[]) { int p, i; int mark_limit; long long sum = 0; memset(bitmask, '\0', sizeof(bitmask)); mark_limit = (int)sqrt(limit); for (p=2 ; p <= mark_limit ; p++) { if (! ( bitmask[p>>3]&(1 << (p&(8-1))) ) ) { /* It is a prime. */ sum += p; for (i=p*p;i<=limit;i+=p) { bitmask[i>>3] |= (1 << (i&(8-1))); } } } for (; p <= limit; p++) { if (! ( bitmask[p>>3]&(1 << (p&(8-1))) ) ) { sum += p; } } printf("%lli\n", sum); return 0; } </pre> This was too fast to measure with 20 runs alone, so 500 runs of it took 15 seconds, two or three orders of magnitude faster than the fastest Haskell or Perl versions. But naturally, we can apply the halving paradigm there too: <pre> #include <string.h> #include <math.h> #include <stdint.h> #include <stdio.h> #define limit 2000000 int8_t bitmask[(limit+1)/8/2]; int main(int argc, char * argv[]) { int half, p, i; int half_limit; int loop_to; long long sum = 0 + 2; memset(bitmask, '\0', sizeof(bitmask)); loop_to=(((int)(sqrt(limit)))>>1); half_limit = (limit-1)>>1; for (half=1 ; half <= loop_to ; half++) { if (! ( bitmask[half>>3]&(1 << (half&(8-1))) ) ) { /* It is a prime. */ p = (half << 1)+1; sum += p; for (i = ((p*p)>>1) ; i < half_limit ; i+=p ) { bitmask[i>>3] |= (1 << (i&(8-1))); } } } for( ; half < half_limit ; half++) { if (! ( bitmask[half>>3]&(1 << (half&(8-1))) ) ) { sum += (half<<1)+1; } } printf("%lli\n", sum); return 0; } </pre> 500 runs of it take 10 wallclock seconds - 54.35 times per second, and 50% better than the previous C version. And I still haven't applied platform-specific gcc optimisations. I should also note that the executables generated by ghc are extremely large in comparison to their C ones: <pre> $ ls -l c_* haskell_* -rwxr-xr-x 1 shlomi shlomi 6082 2009-12-04 07:46 c_mine -rwxr-xr-x 1 shlomi shlomi 6103 2009-12-04 07:46 c_mine_half -rwxr-xr-x 1 shlomi shlomi 6092 2009-12-04 07:46 c_mine_micro_opt -rwxr-xr-x 1 shlomi shlomi 796825 2009-12-04 07:56 haskell_mine -rwxr-xr-x 1 shlomi shlomi 571717 2009-12-04 07:46 haskell_zeroth </pre> <tt>c_mine_half</tt> is less than 1% the size of <tt>haskell_mine</tt> (and runs faster). When talking about this to other people, they said that Haskell has a very optimised primes sequence generator, which I can try using (which should be over 3 times as fast), and that it has two kinds of integers, which the other type is faster, and that it has a better way to emulate assignment. But the bottom line is that the naïve and intuitive way to write such programs in Haskell is under-performant, even in comparison to Perl, and 100 or 1,000 times as much in comparison to C. I've written this as a separate post and not as a comment to the original blog post because I'm very limited with the markup in the commenting there (I'm going to post a comment there with a link to this blog post, though). I should note that you can find all the code I mentioned inside <a href="http://github.com/shlomif/Project-Euler-Problem-10-Comparison">a dedicated Github repository</a>, and you can experiment with it further. Shlomi Fish ( shlomif@iglu.org.il ) problem summary primes project optimize optimise optimization c haskell perl 10 speed optimisation euler http://community.livejournal.com/shlomif_tech/44006.html http://community.livejournal.com/shlomif_tech/44006.html Sat, 30 Jan 2010 12:50:07 GMT Freecell Solver 2.40.0 was Released http://fc-solve.blogspot.com/2010/01/freecell-solver-2400-was-released.html <a href="http://fc-solve.berlios.de/">Freecell Solver</a> version 2.40.0 has been released. It is available in the form of a source tarball from <a href="http://fc-solve.berlios.de/download.html">the download page</a>. This release contains a fix to a string overflow with processing the command line arguments, and an optimised command-line preset that can be invoked as <tt>-l blue-yonder</tt> (or <tt>-l by</tt> for short) that solves the Microsoft 32,000 deals in under 100 seconds on a Pentium 4 2.4GHz machine. It also contains some more minor changes: there is now a <tt>Scan:</tt> header with the name of the current soft thread, when debugging under <tt>-s -i</tt>, an off-by-1 iterations count was fixed and the iteration handling callback is now applied globally to all the instances. Finally, <tt>make_pysol_freecell_board.py</tt> has support for generating PySol's and PySolFC's <a href="http://en.wikipedia.org/wiki/Black_Hole_%28solitaire%29">"Black Hole" Solitaire</a> deals, intended for my newly released <a href="http://www.shlomifish.org/open-source/projects/black-hole-solitaire-solver/">Black Hole Solitaire Solver</a>. Enjoy! <div class="blogger-post-footer"><img width='1' height='1' src='https://blogger.googleusercontent.com/tracker/96099636168698788-8433576534342073452?l=fc-solve.blogspot.com' alt='' /></div> Shlomi Fish ( shlomif@iglu.org.il ) speed release optimization security minor solving pysolfc overflow version freecell 2.40.0 pysol buffer solver asciidoc tag:blogger.com,1999:blog-96099636168698788.post-8433576534342073452 Wed, 27 Jan 2010 20:33:00 +0000 2010-01-27T12:34:13.248-08:00 XML-Grammar-Fortune version 0.0200 Was Released http://community.livejournal.com/shlomif_tech/43620.html I recently released version 0.0200 of the Perl 5 CPAN Module <a href="http://web-cpan.berlios.de/modules/XML-Grammar-Fortune/">XML-Grammar-Fortune</a>. XML-Grammar-Fortune is a module to maintain a collection of Unix-like fortune cookies as XML and convert them to XHTML and plain-text. New in this release are a heavily fixed and enhanced code for rendering the XML to plaintext and a more enhanced META.yml file. XML-Grammar-Fortune was successfully utilised for <a href="http://www.shlomifish.org/humour/fortunes/">my collection of Fortune Cookies</a>. Shlomi Fish ( shlomif@iglu.org.il ) release unix text fortune grammar hacktivity xml-grammar-fortune perl xml fortunes cpan http://community.livejournal.com/shlomif_tech/43620.html http://community.livejournal.com/shlomif_tech/43620.html Mon, 25 Jan 2010 15:49:53 GMT The CPAN Dependencies of Website Meta Language http://community.livejournal.com/shlomif_tech/43345.html In the past few days I resumed work on converting the <a href="http://thewml.org/">Website Meta Language</a> (WML) build-system from the GNU Autotools (also known as "GNU Autohell") to <a href="http://www.cmake.org/">CMake</a>. This involved implementing a lot of the existing makefile and autoconf logic in CMake and Perl. Yesterday, I converted the "wml_common" sub-directory which turned out to contain some ancient versions of CPAN distributions with a forced-upon way to install under the WML prefix. I decided to forego installing all of them and just check for their existence in the perl version that wml uses using some WML and "perl -MMyModule" code. I was able to find all the needed modules in the Mandriva repository, except one - <a href="http://search.cpan.org/dist/File-PathConvert/">File-PathConvert</a> , which I've never heard about. I checked it on CPAN, and saw that it was deprecated and superseded by File::Spec and Cwd. So I grepped the tree for its use and found it was used only in one Perl program which I quickly converted to File::Spec and Cwd. Now, the future build system of Website Meta Language will be simpler and it won't use deprecated modules. Oh! And <a href="http://blog.mozilla.com/blog/2010/01/21/firefox-3-6-release/">Firefox 3.6 (Gamma) is out</a>! I've been using the betas and RCs on Mandriva Cooker for some time now, and now I've upgraded to the stable version and it's still great. Shlomi Fish ( shlomif@iglu.org.il ) eumm wml language cmake perl firefox gnu autohell website meta cpan http://community.livejournal.com/shlomif_tech/43345.html http://community.livejournal.com/shlomif_tech/43345.html Fri, 22 Jan 2010 19:22:40 GMT Perl: Using Vim's snipMate for Perl 5 snippets http://community.livejournal.com/shlomif_tech/43262.html I noticed that there was some Perl 5 code that I had to type or copy-and-paste again and again on many occasions. So I decided to find a way to put it in one place and then recall it. I thought of writing it myself, but then recalled the <a href="http://www.vim.org/scripts/script.php?script_id=2540">snipMate snippets extension for Vim</a> (which <a href="http://www.catonmat.net/blog/vim-plugins-snipmate-vim/">Peteris Krumins covered in a blog post</a>), which allows that. After a little reading, I was able to prepare the snippets. I placed the following under <tt>~/.vim/snippets/perl/_slurp.snippet</tt> : <pre> sub _slurp { my $filename = shift; open my $in, "<", $filename or die "Cannot open '$filename' for slurping - $!"; local $/; my $contents = <$in>; close($in); return $contents; } </pre> And now I can type "_slurp<TAB>" to recall it. The existing "slurp" snippet in <tt>~/.vim/snippets/perl.snippets</tt> is quite evil, with an ugly inline line, typeglobs, and a two args open without a die statement. Thanks, but no thanks. Afterwards I added the following in <tt>~/.vim/snippets/perl/_ltestb.snippet</tt> <pre> local $Test::Builder::Level = $Test::Builder::Level + 1; </pre> And now I can type "_ltestb" (short for "local Test::Builder") and put this line there to create my own custom Test::More/Test::Builder tests. Thus, using snipMate, you can create your own short-hands for commonly-used snippets like that. Shlomi Fish ( shlomif@iglu.org.il ) test tip local slurp perl snipmate vim tip builder snippets vim http://community.livejournal.com/shlomif_tech/43262.html http://community.livejournal.com/shlomif_tech/43262.html Thu, 21 Jan 2010 16:26:13 GMT Report on the TelFOSS "Moose for Beginners" Meeting http://community.livejournal.com/shlomif_tech/42895.html Last Sunday, <a href="http://tel.foss.org.il/">Tel Aviv Open Source club (TelFOSS)</a> met to listen to Sawyer's presentation about "Moose for Beginners". Despite the fact that the meeting was well-publicised, very few people came. Sawyer did not prepare a lot of slides and since there were not many questions he finished quickly. After all that, he went to demonstrate how to implement several use cases that the audience has proposed as Moose programs, and despite that we still finished early. I asked about implementing an "XML parser", but in fact I wanted to implement a processor for a certain XML grammar. In any case, he misunderstood and told me to use Parse-RecDescent for that. I've grown to dislike P-RD for most stuff though. In any case, I'm getting tired of the low attendance in TelFOSS meetings. We do so much work in organising and publicising the meetings, and most people don't bother to come. Are the people too busy? Did they run out of steam? Why don't they come? This makes me frustrating and unwilling to further organise the TelFOSS meetings. Shlomi Fish ( shlomif@iglu.org.il ) telux telfoss meeting open perl club sawyer moose tel aviv source report http://community.livejournal.com/shlomif_tech/42895.html http://community.livejournal.com/shlomif_tech/42895.html Wed, 20 Jan 2010 12:35:04 GMT The Quest for a Good Window Manager / Desktop Environment http://community.livejournal.com/shlomif_tech/42522.html I'm using Linux (Mandriva Cooker) and have been routinely starting many window managers and desktop environments as the need arises. Lately, I've been working a lot with KDE 4 and I happen to agree with <a href="http://www.osnews.com/story/22662/Your_KDE4_Experiences">OSNews.com's assertion that it is slow, sluggish, and has frequent random crashes</a>. I'd like to move to something more usable until the KDE 4 programmers get their act together. Now, I've traditionally used <a href="http://www.icewm.org/">IceWM</a> but it no longer cuts it for me because it does not have a different wallpaper for every virtual desktop. XFCE fails on the same premise. Enlightenment 17 has a different wallpaper per virtual desktop, but its task bar is dysfunctional, if not completely non-existent, and it has many other very annoying quirks. GNOME is just plain annoying (and doesn't have a different-wallpaper-per-virtual-desktop either), and KDE 3 is no longer present on Mandriva. LXDE does not have it either. Now I dislike the concept of <a href="http://en.wikipedia.org/wiki/Tiling_window_manager">tiling window manangers</a>, of which there has been an inflation, recently, and I am looking for a window manager that has stacking. And like I said, it also needs to have a usable task bar, a different wallpaper per virtual desktop and be responsive and stable. Do you have any recommendations? Shlomi Fish ( shlomif@iglu.org.il ) window desktop tiling stacking unix kde 4 x11 wallpaper environment virtual icewm x xfce manager windows workspace linux http://community.livejournal.com/shlomif_tech/42522.html http://community.livejournal.com/shlomif_tech/42522.html Mon, 18 Jan 2010 19:49:59 GMT