|
Re: Version Control is Undo
|
Posted: Dec 30, 2008 6:42 AM
|
|
> wow now that i’ve actually skimmed over the other posts i > must say this discussion has gone haywire. it started out > with a discussion of version control, but has turned into > a little xml war. > > i’ll say, for all that xml has done good, it is still 80% > B.S.---things you don’t need, things you don’t want, and > things you can hardly use. the single most important thing > in IT is data, the second important thing is the structure > of that data, and the third, to process / manipulate / > transform that data. > > xml does nothing that is easily usable to approach the > first problem---in xml, you can’t even tell a number from > a text, or a boolean (there is xml schema, and there is > DTDs, but both are atrocious solutions and, for me, > necessary evils that largely fall into the 80% B.S. > category. DTDs are not even formulated in XML). > > xml does astoundingly little to approach the second most > important point in IT---the structure of data. sure, you > can put tags inside of tags, but you are still burdened > with all that attribute-vs-tag content B.S., plus you > still don’t know whether > (foo)(bar)42(/bar)(baz)108(/baz)(/foo) is a list of values > or a mapping of values. people, JSON does get the first > point right (for five most important things: number, text, > null, true, false), and it gets the second point right > (with lists and mappings). AND it does so with 1%---ONE > MEAGER FUCKING PERCENT---of all the complexities of XML > (or, god beware, SGML). (disclaimer: yes, i know it would > be hard to use JSON as a text markup technology, but > frankly some of the most annoying B.S. that you have to > deal with when parsing XML-as-data-structure-language > comes down on you because of mixed content).
I couldn't agree with you more! Referring to XML was an example of a solution in the right direction. JSON is a more suitable solution to the problems you mention.
To come to the correct path in the discussion, a format like JSON (or XML, whatever) which is structured and provides a set of primitive types would be ideal for version control and undo.
> > lastly, XML itself does nothing for you when it comes to > processing data. only tools tailored to handle its immense > complexities---DOM, SAX, whathaveyou---can do that for > you. that is all fine if it wasn’t for the fact that the > complexities of XML do shine through in almost every tool > you can use in the most obnoxious of ways. heck, when i > retrieve a JSON structure over the wire via an ajax HTTP > request, i already have e.g. a NATIVE javascript mapping > of names to NATIVE booleans, strings, numbers---am i > supposed to implement all of XML schema in javascript so i > can have numbers in the result set? say WHAT??
This problem does not diminish the use of XML in any way, and it's irrelevant to XML though.
> > it took people like john resig, of jQuery fame, to show > the world that you can, in 80% of the time, get away with > `$('#foo')`, `$('.bar')` and friends. in may experience, > that is also 80% of what you ever wanted to do with XML. > the simplicity, beauty, and usefulness of his design > stands in a shockingly stark contrast to the immense heap > of specs that have gone into making XML the bloat it is. > which is sad, as it looked like a good idea, years ago.
The point of this discussion is how version control and undo could be merged, and using a structured format would help.
|
|