Re: RFS: yajl
> On Tuesday 04 August 2009 12:07:11 pm George Danchev wrote:
> > I haven't review the package, but it would be very nice if anyone can
> > compare that to the json-c project , which looks quite serious to
> > me, at least as briefly scanning through the code. We have JSON
> > implementations in almost any language already available in Debian
> > (including a C++ one, which is nice), but not for C. I haven't used
> > neither of these, but would appreciate a review by someone who has
> > used any of these in their apps.
> >  http://oss.metaparadigm.com/json-c/
> > (documentation is scarce)
> I haven't done a comparison either, but it looks like yajl was created
> partly out of frustration with json-c.
> As for applications, argyll includes libyajl in its private
> libraries, and when lastfm-desktop is released it will most likely
> have it among its build-depends. gdal embeds json-c.
>  http://trickyco.de/2009/05/parse-json-in-ruby-in-14-the-time-of-yaml/
>  http://packages.qa.debian.org/a/argyll.html
>  http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=537200
>  http://packages.qa.debian.org/g/gdal.html
Aha, it is the stream parsing and data representation independence which makes
the difference. It looks to me that as the time goes by the applications will
slowly migrate to yajl if they want to benefit from the above two key features.
Json-c would need a major restructure to provide these, which is highly
The package looks solid, everything is put in place and I was unable to find
any flaws other than boring nitpicking:
* the short description of the libyajl1-dbg should read `... - debugging
symbols', instead of ` - library documentation'. Probably a copy/paste ;-)
* examples would help users to bootstrap more easily (optional)
* a watch file would help package monitoring
* send man pages upstream along with the patch of `PIC does not make sense for
static libs' ;-)
* Why the integer test fails on 64-bit machines? Aligning issues?
* I imagine that at this stage the API is stable enough and drifts should be
highly unlikely, is that assumption correct for yajl?
Let's give some days to the rest interested to look at that library package,
to see if they can find any flaws. Then we can upload.
pub 4096R/0E4BD0AB 2003-03-18 <people.fccf.net/danchev/key pgp.mit.edu>