[Csync2] old patches and such

[Art -kwaak- van Breemen]

On Tue, Dec 15, 2009 at 11:38:44AM +0100, Lars Ellenberg wrote:
> And I had a look at what you do in csdb.
> 
> what I dislike is CamelCase and camelCaseReverse_Polish_Notation_ ;)
> it does not fit into the existing csync2 coding style too well.

Well, it doesn't fit anything I guess ;-). I've been programming
OO too much in bash(!). It really helps out to spell out the
arguments, and I kinda borrowed that from Smalltalk ;-).
( doThis: with: gets converted to doThis:With: as a single
literal).

> and I don't see the point in passing around a global variable either
> (limits).

Hmmmm, limits was a real hack.

> can you verify my commits so far?
> could you pipe the csdb* stuff through a camel case remover,
> and indent?
> if you don't have the time to do that,
> would you be ok if I do that instead?

I have to be ok with that because you are upstream 8-D.

> does your "experimental" ipc and csdb stuff work for you now reliably?
The ipc stuff is a small library I used in another project. The
csdb stuff as it is now is in "production": it syncs large
quantities of data.

> do you have some sort of "regression tests", or do you just
> do some by hand plausibility check and then go live in prod?
I am affraid I do the latter.

A good regression test scheme would be nice. It won't find all
bugs though. I just found a bug (thanks to csdbd ;-) ), or more a
problem with sqlite3 where it bails out on a
SQLITE3_SCHEMA(change). 

This actually only happens on our freeze3 systems, which is a
debian unstable release between sarge and etch.

BTW: I did some new fixes, so the begin/commit also works in
csdbd mode.
But I first have to check your commits, and fix the camels.