CPU hogging in the current master

[Linus Torvalds]

On Thu, Sep 26, 2013 at 2:05 PM, Linus Torvalds
<torvalds at linux-foundation.org> wrote:
>
> I may do a first private version for myself that only saves in a
> non-xml format..

Hmm. So I haven't finalized this, but I have a trial format that is
basically our simplified xml, but without the xml crap. The plan is
that I can feed it to our stupid xml parser by "faking" the xml part,
rather than actually using that horrible libxml. And my stupid fake
parser would be purely line-based.

No begin/end tags.. No quoting, because there are no special
characters except for a field that can have multiple lines (ie notes),
and then that is defined by the next line having the first non-space
character being '|'. I guess you can call that "quoting" (technically
it's an odd quoting of the newline).

I might end up quoting the quote character just to allow the nice
"let's put dive computer names etc in quotes", but I haven't decided
yet. I might just give those kinds of tags lines of their own.

And samples are special. Drop the 'time' and 'depth' nametags
entirely, since they always have to be there.

Attached is what my last dive in Florida looks like with the
(partially) done format I have in mind. Not final. Design goal: still
be very much human-readable, while being denser than the xml (by
dropping all the crap) and still having the tags so that we can allow
basically arbitrary extensions.

Is anybody so married to xml that they hate the idea of having a
native subsurface file format that is basically de-crapified xml with
some hardcoded formatting knowledge (ie that format for a sample line
etc)?

                         Linus
-------------- next part --------------
A non-text attachment was scrubbed...
Name: one-dive.sbs
Type: application/octet-stream
Size: 31374 bytes
Desc: not available
URL: <http://lists.hohndel.org/pipermail/subsurface/attachments/20130926/c01f73d4/attachment-0001.obj>