Re: [xml] pretty-print & format

[Daniel Veillard <veillard redhat com>]

On Fri, Aug 02, 2013 at 10:07:58AM -0400, Liam R E Quin wrote:
> On Fri, 2013-08-02 at 14:53 +0100, Alex Bligh wrote:
> > On 2 Aug 2013, at 14:38, Liam R E Quin wrote:
> >
> > > On Fri, 2013-08-02 at 11:33 +0300, Afanasiy Fet wrote:
> > >
> > > > what may help reduce network load.
> > >
> > > Efficient XML Interchange (EXI) is a W3C spec for XML-specific
> > > compression that may be of interest, depending on what you're doing.
> >
> >
> > gzip works well and is built into libxml (at least on the reader side I think)
>
> "well" is subjective of course, but yes, gzip can be effective as a
> first cut. EXI has the theoretical advantage of transferring the
> abstract tree, not the syntax, so that the reader doesn't need to
> decompress the input and then parse pointy brackets, but can
> reconstitute an in-memory tree directly. It can also make use of a
> shared schema to get further reductions in bandwidth use. These
> advantages can be especially significant on mobile, low-performance or
> fixed-memory devices.

  My stupid cell phone has more memory and horsepower than my workstation
when I started working on libxml. I don't see any problem on the client
side myself.
  I have more problem with my browser exploding with the complexity of
what it needs to handle, resulting in having to kill it off on a regular
basis, than it having a CPU probem parsing the data. I will argue that
what we need is a simpler stack, not nano-optimizations for the few
parts which are actually staight and rather reliable. YMMV !

Daniel

-- 
Daniel Veillard      | Open Source and Standards, Red Hat
veillard redhat com  | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | virtualization library  http://libvirt.org/