←back to thread

What If OpenDocument Used SQLite?

(www.sqlite.org)
275 points whatisabcdefgh | 2 comments | 04 Sep 25 21:36 UTC | HN request time: 0.731s | source
Show context
librasteve ◴[04 Sep 25 22:51 UTC] No.45133139[source]▶
wouldn’t an XML database be easier?
replies(7): >>45133167 #>>45133236 #>>45133299 #>>45133364 #>>45134339 #>>45134462 #>>45136358 #
duskwuff ◴[04 Sep 25 23:23 UTC] No.45133364[source]▶
You can't* index into XML. You have to read through the whole document until you get to the part you want.

*: without adding an index of your own, at which point it isn't really XML anymore, it's some kind of homebrew XML-based archive format.

replies(1): >>45138248 #
1. Mikhail_Edoshin ◴[05 Sep 25 13:20 UTC] No.45138248[source]▶
This applies to any secondary index. The data themselves can only be ordered by a single criteria. It may be a meaningful one, but I guess in most cases it is merely the internal ID, which means you will have to scan the whole table too.

XML was meant for documents so in most cases the sequence of elements is given. But technically if I compose XML myself I can lay it out the way I want and thus can have it sorted too. This means it will be directly searchable without an index: read a bit at the middle, find an element name, see where we are, choose head or tail, repeat.

replies(1): >>45143694 #
2. duskwuff ◴[05 Sep 25 21:13 UTC] No.45143694[source]▶
Blindly seeking into XML data is a risky, error-prone approach. It's not impossible to do, but doing it correctly is difficult - even if the tags you're looking for are unique, there are a lot of messy edge cases involving comments and <!CDATA> blocks.