today one my most admired colleagues had his last day. A big gap is left in the company and also for me as he was one of the people it was nice to learn from. Not only technically and conceptionally but also on a project management level he was not even a real specialist in. Really sad to see one of the best colleagues leaving
June 28, 2006
June 27, 2006
A cool forms generating Ajax Web application. Really nice…
Does it really pay off to build your own stuff now?
June 24, 2006
Just found the quite interesting Websites as Graphs via a review in c’t magazine. The homepage of this site looks like this:
June 20, 2006
I never really thought about which characters are allowed in XML. I vaguely knew NUL was not but that was all I cared about maybe because I never had a problem (I mostly do XSLT on given XML so no real problem there).
At work we now had a problem with “XML” documents (built with dom4j) which contained characters #28 and #31 probably by copy/pasting text from PDF documents in the XML generating application (and therefor not really XML at all). dom4j never complained generating the XML but of course complained when parsing the resulting document “thing” again as most character between #00 and #32 (+ #127 but except #09 and #13 (which I used in XSLT several times to generate nicer looking text output)) are not allowed in XML at all.
Until now I never had these problems maybe because I used libs like XOM which are smart enough to know when building the XML that these characters are not valid. To be sure I looked in my XML in a Nutshell and found the above information.
I told my coworkers better not to use dom4j some time ago for several problems we encountered in the last year. So another problem added to that pile.
BTW I re-read an interview with Elliotte Rusty Harold on Artima which exactly explained this problem. After that I did a quick test with the simple minidom implementation for Python. Same problem there, generating the XML goes though but results in a broken document which is no longer parsable.
Got to be aware of these problems, luckily most of the time I vaguely remember something like the above interview
I guess I try it out on the next project that comes up…
June 18, 2006
I needed a cross browser way of clearing floats without the usual markup. There are a lot of hacks on the net to find including the ones explained in the nice book “CSS hacks”. Sadly this hack which I used until now does not work in IE7beta2 so I did a bit more googling and found a solution. The problem in IE7 is its very different rendering of HTML and XHTML. So I put all hacks together in 2 pages – one for HTML and one for XHTML (very important to distinguish!)- to test it: CSS clearfix without additional markup. I have tested Firefox 1.5, Opera 8.5, IE 5.01, 5.5, 6 and also IE7beta2 which all work on the test pages. I still need to test Safari (and maybe IE Mac which probably needs another hack added) but have no Mac at home so it has to wait a bit…
Of course there are better sources for this kind of hack but my aim was just to compile a working example which can be used as it is (for my own work too
If you find any problems let me know.
Tests on Safari releaved some more problems which now should be fixed and the examples should work additionally in Camino, Opera8Mac, IE5Mac and Safari.
I know IE preferred usage would be conditional comments but as these require lots of different stylesheets (or at least style sections). I prefer all stuff in only ONE place so still use CSS hacks. It gets rather tedious the more browsers you are targetting but it still seems manageble. Especially XHTML seems the preferred markup for CSS hacks apart from its other advantages (parsable etc)…
June 15, 2006
About a year ago (after I got a new machine I tried to install MySQL 4 after having used v3 for quite some time. I never got it to work with PHP 4 and the current phpmyadmin tool so gave up at that time and installed MySQL 3.23 again.
But I noticed that most providers are using MySQL 4 now (even the cheaper ones some of the sites I sometimes need to update run on) which causes trouble if I want to import the online v4 dump into my v3 database. So up to a new try…
Installation of MySQL itself was not the problem but phpmyadmin did not want to talk to it. At least not with the PHP version I use which still is v4 as most (simple) websites I need to work on still use v4. After some googling I found that the reason is a changed password mechanism. No problem if you are content to use no password at all which is not an option even for my local development machine. So I did a longer googling session and found these forum entries which describe a solution. It is rather complicated but does work after several iterations…
So now all is working again, I even installed the current PHP 4 release. Importing the database data was not too difficult although I had some tries again as some sites still use a latin1 encoding while other newer sites I built all use UTF-8 encoding. I guess the best would be to work only in UTF-8, at least in the sites I “control” myself.
Nevertheless the whole procedure took way too long for a technology I only use because I need to (reasons are availability and price or deployed systems) and not because I like to.
June 6, 2006
Today I recognized the batchvalidator script I wrote some years ago (and have not used for quite a while) was missing a vital option namely to validate dynamic files from a local server which my not be accessible by the online W3C validator. So a quick fix and version 1.6 was done.
During testing I found my own site does not actually validate fully, ohouu – how embarrassing…
I fixed some errors but some others have to wait until I have more time. I guess I should use my own tool more frequently…