Re: [Evolution-hackers] pvl_list structure in libical

[Harish Krishnaswamy <kharish novell com>]

Hi Mathew,


On Fri, 2006-07-14 at 17:06 -0400, Matthew Barnes wrote:
> Hi all, I've recently started maintaining Evolution for Red Hat.
> 
<snip>
> The source code is dated November 1995 and looks to be isolated to
> libical.  It's basically a linked-list of void pointers, and I don't see
> anything useful here that GList doesn't already provide.  Plus, the
> utilization of GSlice for node allocations -- rather than raw
> malloc()/free() calls as pvl_list uses -- might yield a small
> performance improvement.
> 
> I'm tempted to purge it from libical and replace it with a GList
> structure, but first I wanted to check with the upstream maintainers to
> see if they'd accept such a patch.  Are there other issues here that I'm
> not considering, such as API stability?


The libical module inside EDS is a source snapshot obtained from the
upstream few years ago and has remained as such ever since, save a few
functionality fixes and added functions.

Much of e-d-s clients (incl. Evolution) use the ECal APIs that wrap
around libical functionality though there are many places in Evolution
that still invoke libical functions directly.

It has been our direction to code the new functionality in ECal and
slowly ease out libical calls from Evolution in favor of ECal.

Yes. We would be interested in moving towards GSlice based memory
handling replacing icalmemory and use of G(S)List. It has been the
direction we have wanted to move for sometime but has been in the
back-burner in favor of stability fixes and feature additions.

I would be happy to work with you on this .


-Harish