it's more fun to compute.

i am always and ever amazed how my coding skills are so incompatible to the act of coding itself. it's not only that i make shitty software, that bugs get in the way. it's moreover about how i gawkily dance around trying to fix them.

example: after almost two lightyears i once again started to write a helma app working with the internal database. you've got to know that helma's internal database has undergone drastic changes: it's now completely xml-based which sounds really cool and that is what it really is like.

i quickly scribbled the first files to enable my application setting up a simple website providing a document-based infrastructure similar to the old helma.org. ie., i was able to create, edit and display documents mounted to a custom url.

everything went fine so far until i restarted the application for the first time: duh! all data was gone.

this was the initial situation i started to find a bug in my code. i reduced everything down to one script creating a dummy document and kept myself watching the output in the xml database. still the data loss was 100%.

no bug in my code, i thought. so could it be that helma...? maybe the internal database does not work the way i was used to (lightyears, y'know)...

so i started to doubt: the accessname property of the document collection – throw it out! the use of the internal _children property – throw it out!. using type.properties: throw it out!

all without success.

my little dance already had become an eurythmic exercise, a seriously ridicilous and desperate effort to catch an invisible needle in a just as invisible haystack.

but then suddenly it dawned on me and i renamed the prototype called "document" (which is one of the reserved javascript keywords, even on the server side) to "doc".

from then on, everything went fine again. well, except that i was not able to create a persistent subcollection in root called "people", yet. could it be that helma perhaps this time...?