http://www.neverpanic.de/blog/ CoffeeBreak, the neverpanic.de blog en Clemens Lang Creative Commons Attribution-NonCommercial-ShareAlike 2.0 Germany License 2011-08-10T23:00:34+00:00 http://www.neverpanic.de/blog/single/geshify-has-moved/ http://www.neverpanic.de/blog/single/geshify-has-moved/#When:16:52:55Z My syntax highlighting extension for ExpressionEngine has moved to a new home: http://geshify.com/ Web Development 2009-04-02T16:52:55+00:00 http://www.neverpanic.de/blog/single/geshify-update/ http://www.neverpanic.de/blog/single/geshify-update/#When:16:05:47Z GeSHify, the syntax highlighting extension for ExpressionEngine has been updated. I updated my ExpressionEngine extension GeSHify. New features are a French translation (thanks to Fabien Amann) and a German translation (which I did myself). Packaged with GeSHify 0.3.6.1 are GeSHi 1.0.7.22 and GeSHi 1.1.2alpha4dev (both checked out from the release branches of GeSHi’s SourceForge.net Subversion repository). Since I switched from getting the releases of GeSHi manually to using svn:externals definitions I no longer have to download the GeSHi releases manually — they will be pulled from the SourceForge Subversion servers on every update. Using this technique it is technically possible to automate building new GeSHify packages as soon as a new GeSHi version is released (and I will probably set this up soon). Proceed to download. Web Development 2008-08-05T16:05:47+00:00 http://www.neverpanic.de/blog/single/i-took-it-and-so-should-you/ http://www.neverpanic.de/blog/single/i-took-it-and-so-should-you/#When:15:14:57Z A List Apart has published the 2008 survey for people who make websites. If you work in the web business, you should go ahead and take the survey as well. Web Development, Personal 2008-07-29T15:14:57+00:00 http://www.neverpanic.de/blog/single/thunderbird-imap-push-email/ http://www.neverpanic.de/blog/single/thunderbird-imap-push-email/#When:10:18:49Z Using Thunderbird as the e-mail client of your choice and the IMAP protocol to fetch and manage your mail? RFC 2177 defines IDLE, an extension to the IMAP protocol allowing mail clients to change into an idle mode while keeping the TCP-connection open. This allows the servers to notify the clients of new mails on arrival (the so-called push e-mail). Not all e-mail providers support IMAP IDLE, though – you should check whether your mail provider supports it by opening a telnet connection to the IMAP server (standard port for IMAP is 143) and sending 001 capability. If the answer contains „IDLE” your server supports IMAP IDLE. GMail for example answers * CAPABILITY IMAP4rev1 UNSELECT IDLE NAMESPACE QUOTA XLIST CHILDREN XYZZY 001 OK Thats all she wrote! There's one thing to look out for, though – Thunderbird only sends the IDLE command (effectively enabling push e-mail) if you disable the „check for new mail every nth minute”-option. I could not find any documentation on that feature – however using Help » Mozilla Thunderbird Help causes a 404 File Not Found error for me anyway (using the German version of Thunderbird). Steve Jobs would say: „Exchange for the rest of us” – using GMail, IMAP and Thunderbird. Web Development, Personal 2008-07-12T10:18:49+00:00 http://www.neverpanic.de/blog/single/probably-the-biggest-rss-icon-i-have-ever-seen/ http://www.neverpanic.de/blog/single/probably-the-biggest-rss-icon-i-have-ever-seen/#When:16:07:00Z There’s a certain trend in so-called web 2.0 design to have ever bigger icons for RSS feeds (What are RSS feeds?). Now I recently saw an RSS icon that beats the pants of all other RSS icons I’ve ever seen before: ImageShack’s RSS icon (screenshot): This icon is a whooping 403 per 336 pixels large - that’s 135,408 square pixels! Web Development 2008-06-13T16:07:00+00:00 http://www.neverpanic.de/blog/single/a-rant-on-limited-password-fields/ http://www.neverpanic.de/blog/single/a-rant-on-limited-password-fields/#When:19:58:00Z You register for a new service, type in all your details, choose a password - and after 16 characters the password field stops accepting your input. Whenever I see a password field that’s limited in it’s length I loose some trust in the organization providing the form. Why? Because the only valid reason I see to limit the length of a password is unencrypted or reversible encrypted storage. This requires some insight knowledge on how data is stored in databases: when defining a text field you usually have to choose it’s maximum length - obviously you could just choose whatever maximum length was possible, but that’s not the point of a well designed database. It would take me too long to explain that here, but there are some reasons why you certainly wouldn’t want to define fields longer than needed. Password encryption methods There are several different approaches to password encryption; the first and probably worst method is no encryption. It’s advantages are that you will be able to mail the password out to the user in case he forgets it and that it doesn’t require a lot of storage (estimating the average password length at about 10 chars from personal experience here). However it’s downsides are huge: in case you have some security flaw in your application all passwords will be leaked. The second way is to use some encryption algorithm, e.g. AES or DES (those two might be a bit overkill, but there’s others). Strings encrypted with such an encryption algorithm generally tend to use up a lot more storage than the unencrypted string. You still have the advantage of being able to send the password to the user and if your database is leaked, the passwords will still not be decryptable (as long as the encryption key isn’t leaked). However, there’s a better solution: hashing the password will give you a fixed-length string (which is great if you’re working with databases) from a variable-length string (=password) and it’s non-reversible. Theoretically you’d be able to calculate a so-called “collision”. Because a hash function is mapping an endless amount of strings to a finite amount of hash values there’s multiple strings with the same hash value. However given a certain minimum length of the password (which can be enforced by padding the string with random data you also store, the so-called “salt”) the effort to calculate such a collision is that large that it’s virtually impossible to hit one while your site is online (assuming you to re-setup the system once in 100 years here). Maybe you’d ask how to compare passwords on login if they’re stored using a non-reversible encryption? Actually you wouldn’t. You would instead hash the password the user entered with the salt stored for the user and compare it to the stored encrypted password. That way the unencrypted password is never in the server RAM and not in any log files either. And what if a user forgets his password? Well, instead of mailing him his password, you could just generate a new password, encrypt it, store it into the database and send the new password to the user. Also the hashing method allows you to enter virtually every character regardless of things like the encoding the database uses internally. Because hashing functions work with binary data and usually return a string using 0-9 and a-f (which are always the same in any encoding relying on ANSI) there won’t be any problems with the database encoding to deal with either. So, why is there still services not allowing long passwords and all characters in passwords? Web Development 2008-05-27T19:58:00+00:00 http://www.neverpanic.de/blog/single/9-slice-scaling-in-html/ http://www.neverpanic.de/blog/single/9-slice-scaling-in-html/#When:14:59:19Z 9-slice scaling is a technique from Adobe’s Flash product line and has been adapted into Fireworks. It is used to scale certain parts of a symbol different from others when resizing the symbol (i.e. not scaling the corners at all) to make corners and borders look the same at every size of the symbol. With a little CSS and HTML this can be done in browsers, too. The Adobe Fireworks help has a good article explaining 9-slice scaling. Now imagine you’re designing a liquid width layout for a website - wouldn’t it just be great if it would scale as it was 9-sliced? Here’s how: Slicing First you need one image for each of the outer areas in the 9-slice grid. That’s top left, top center, top right, left center, right center, bottom left, bottom center and bottom right. We don’t need an image for the center, since it’s filled with one solid color in this example and we can use CSS for that. Note that all top and all bottom slices must share the same height. The top center and bottom center slices should be 1 pixel wide, since we’re not going to scale them, but repeat them (which is the difference of this technique compared to the original 9-slice scaling). Left and right center slices only need to be 1 pixel high for the reasons just mentioned. After all, the slicing looks like this (forgive me the typo): HTML code Here’s the HTML you’ll need. Read the comments in the code for explanation. <div class="content"> <!-- this is the overall container --> <div class="content-top"> <!-- this contains the top left, top right and top center slices --> <div class="content-top-left"></div> <div class="content-top-right"></div> <!-- notice the top center or top repeat slice needs to be at the end --> <div class="content-top-repeat"></div> </div> <div class="content-repeat"> <!-- this is just a container for the vertically scaling part - you could leave it out, but it seems cleaner to me --> <div class="content-repeat-left"> <div class="content-repeat-right"> <!-- in this case the left center and right center must be nested and must contain the actual content div, so they automatically scale with it --> <div class="content-repeat-center"> <!-- /** this is where you'd place the content **/ --> <div class="content-repeat-clearfix"><!-- this will prevent problems with floated stuff within the content --></div> </div> </div> </div> </div> <div class="content-bottom"> <!-- this contains the bottom left, bottom right and bottom center slices --> <div class="content-bottom-left"></div> <div class="content-bottom-right"></div> <!-- as above the bottom center div must be after the bottom left and right divs --> <div class="content-bottom-repeat"></div> </div></div>  Web Development 2008-05-15T14:59:19+00:00 http://www.neverpanic.de/blog/single/up-or-out-employees-in-it-companies-quitting/ http://www.neverpanic.de/blog/single/up-or-out-employees-in-it-companies-quitting/#When:19:14:00Z I just read an article in Alex’ Soapbox on IT professionals and their motives to quit a job at a certain company. Having just finished a 6-month-internship at a local new media agency this article came right on time. I don’t want to judge neither the article nor the company I’ve been working at, but the theory mentioned in “Up or Out: Solving the IT turnover crisis” is absolutely worth reading if you haven’t already. Personally I think Alex Papadimoulis’ findings are a bit too black-and-white (as he admits near the end of the article: “Obviously, this article has painted some incredibly broad strokes, the largest being the stark dichotomy between skilled and unskilled developers, and the lack of distinction between organizations.”), but they are true nonetheless. Web Development 2008-04-30T19:14:00+00:00 http://www.neverpanic.de/blog/single/the-smaller-the-error-the-bigger-the-bug/ http://www.neverpanic.de/blog/single/the-smaller-the-error-the-bigger-the-bug/#When:10:05:52Z Bug tracking in software development can be a pain - especially when there doesn’t seem to be any logic behind the bug. So we had that navigation tree at work consisting out of two category levels, for simplicity let’s call them Level-1-Category and Level-2-Category. Level-1-Categories could contain other Level-1-Categories and Level-2-Categories, while Level-2-Categories could only contain Level-2-Categories. The navigation was built, caching was implemented, everything seemed fine from a logic viewpoint, however it behaved weird where it would sometimes display the Level-2-Categories in a particular Level-1-Category and sometimes not or in different order. When we finally found the bug (after hours of debugging) the scales fell off our eyes: <?php $cacheID = md5($level1Cat->id.$level2Cat->id); ?> Now imagine what happened when you had a $level1Cat with ID 1 and a $level2Cat with ID 12, that would write it’s cache to md5(‘112’); - and trying to open a $level1Cat with ID 11 and a $level2Cat with ID 2 it would read the cache md5(‘112’);. Unfortunately these classes did have the same interface, so no errors were thrown. Changing the code to <?php $cacheID = md5($level1Cat->id.'-'.$level2Cat->id); ?> resolved the bug and finally fired the closing time-event for us. Web Development 2008-04-26T10:05:52+00:00 http://www.neverpanic.de/blog/single/i-have-a-dream/ http://www.neverpanic.de/blog/single/i-have-a-dream/#When:21:16:00Z Ads - That’s how you make money on the Internet. And since everybody wants money ads are spilled all over the Internet. What if… Semantics is the reason why HTML exists: there are elements to tag headlines, tables, lists and countless other types of text to give the text a structure. I know ad providers wouldn’t even consider this, since it would make their ads easily blockable by ad-blockers like Adblock Plus for Firefox, but seeing German sites are required by law to mark advertisements and make them stick out of the text flow why isn’t there an <ad> element (or at least one on the to-do-list for HTML5) to make that tag machine-readable? The only way to make ad providers use such an element would probably to require it by law - and given the expertise the German government has recently shown in laws concerning information technology and the Internet years will be passing by without an <ad> element *sigh* Web Development 2008-03-31T21:16:00+00:00