On Mon, 19 Nov 2012 10:05:19 +0100 Martin Jansa <[email protected]> said:

> On Mon, Nov 19, 2012 at 05:48:35PM +0900, Carsten Haitzler wrote:
> > On Mon, 19 Nov 2012 08:41:15 +0100 Martin Jansa <[email protected]>
> > said:
> > 
> > > On Mon, Nov 19, 2012 at 04:26:30PM +0900, Carsten Haitzler wrote:
> > > > On Mon, 19 Nov 2012 06:28:45 +0100 Vincent Torri
> > > > <[email protected]> said:
> > > > 
> > > > > On Mon, Nov 19, 2012 at 2:06 AM, Carsten Haitzler
> > > > > <[email protected]> wrote:
> > > > > > On Sun, 18 Nov 2012 17:25:13 +0100 Vincent Torri
> > > > > > <[email protected]> said:
> > > > > >
> > > > > >> On Sun, Nov 18, 2012 at 5:16 PM, Andreas Volz <[email protected]>
> > > > > >> wrote:
> > > > > >> > Am Sat, 17 Nov 2012 18:23:02 +0100 schrieb Vincent Torri:
> > > > > >> >
> > > > > >> >> On Sat, Nov 17, 2012 at 5:42 PM, Andreas Volz
> > > > > >> >> <[email protected]> wrote:
> > > > > >> >> > Am Sat, 17 Nov 2012 11:21:58 +0000 schrieb gpl4all:
> > > > > >> >> >
> > > > > >> >> > Interesting!
> > > > > >> >> >
> > > > > >> >> > Does enesim stand in some sort of competition compared to
> > > > > >> >> > part of EFL? Or do they have another mission?
> > > > > >> >>
> > > > > >> >> it's a vector-based rendering library, which could be in
> > > > > >> >> competition more with cairo than with evas. But Enesim could be
> > > > > >> >> used as rendering engine for evas, for example.
> > > > > >> >>
> > > > > >> >> To be more accurate, Enesim is the name of the project. Its
> > > > > >> >> stack is composed of several libraries
> > > > > >> >> Enesim
> > > > > >> >> Emage
> > > > > >> >> Etex
> > > > > >> >> Etch
> > > > > >> >> Ender
> > > > > >> >> Egueb
> > > > > >> >> etc....
> > > > > >> >>
> > > > > >> >> See the wiki in the website for mor einformations.
> > > > > >> >
> > > > > >> > It seems to use at least Eina as base. And then Evas uses it
> > > > > >> > again. Funny...
> > > > > >> >
> > > > > >> > As Eina is now in Efl source together with Evas it's a funny
> > > > > >> > cyclic dependency. :-P
> > > > > >>
> > > > > >> yes. So we can do a bootstrap first. Anyway, I think that this
> > > > > >> bootstrap will be needed for edje, iirc
> > > > > >
> > > > > > thats very different. with edje we can make the bootstrap happen as
> > > > > > part of the efl build, BUT we can't with esvg because it's external.
> > > > > 
> > > > > we can just build eina, then esvg, then the whole stack
> > > > 
> > > > we can't because we dont control the build of esvg. it's not part of the
> > > > build tree.
> > > > 
> > > > what the issue here is, is that packagers what to do this:
> > > > 
> > > > configure XXXXXX
> > > > make
> > > > make install
> > > > package results
> > > > 
> > > > the do not want to (and will avoid like the plague):
> > > > 
> > > > configure XXXX (eina only)
> > > > make
> > > > make install
> > > > package pkg1
> > > > 
> > > > <compile + package depenedncy>
> > > > 
> > > > configure XXXXXX
> > > > make
> > > > make install
> > > > package results
> > > > 
> > > > cross-compile systems aside (and scratchbox fixes the native tools vs
> > > > target tools thing), packagers will complain, moan and avoid efl and e
> > > > if we make them have to package it TWICE - yes. they will insist on
> > > > having to package it twice because their build systems install
> > > > dependencies based on package build depends. that means they have to
> > > > create and initial eina package just to be able to build esvg, then
> > > > just to re-build efl and somehow disable the eina build then (to ensure
> > > > the same eina is used from before) OR.. to replace the previous eina
> > > > which now rsvg was built against.. which they will not like either etc.
> > > 
> > > As meta-efl (layer for OpenEmbedded) maintainer, I fully agree with
> > > this.
> > > 
> > > And it would be much better to disable eina build when building the
> > > rest, because we cannot replace eina in sysroot (it needs only one
> > > "owner" of files staged by build).
> > 
> > well once the efl tree gets to the point of needing to compile elementary,
> > then we need to make it have its own bootstrap phase - compile, but install
> > into some tmpdir in the build dir and run the edje_cc & use the all the
> > libs from there (set PATH, LD_LIBRARY_PATH etc.). yes - i know this doesn't
> > work in pure cross-compile. for a pure cross-compile you'd need to disable
> > the bootstrap via some option (yet to be done) and provide tools like
> > edje_cc and eet in the cross-dev sysroot. then you have to provide much
> > more than eina - u need eina, ecore, embryo, eet, edje... for starters. :)
> > (though as OE does now - it's a minimal build that doesnt need x11 etc.).
> 
> Yes, but those are native deps, which are staged in separate sysroot, 
> easy to define and target recipe then needs just path to correct edje_cc
> in that sysroot etc.

i know - pure cross-compile needs them staged in the native format to be able
to be run first. so you need a 2 stage compile process for efl regardless.

> But for esvg if I understood it correctly you need target eina. 
> So we would have to add eina-initial recipe which builds only eina from
> efl tree in 1st step and stage it in some eina-initial prefix, 
> then build esvg (and hack configure to use eina only from eina-initial
> prefix) and then build complete efl.

but useless for others because you are already doing a bootstrap because you
have a cross-dev setup. they are not. you basically force a multi-stage
package build process on EVERYONE to be figured out themselvesas to hoe to do
it. this is just not acceptable. ignore the esvg case - the nasty one is
edje_cc, and we can solve that internally by doing stages within the build -
much like gcc does. esvg, as long as it remains out of tree is an "unsolvable"
as it stands.

> Or we can "bundle" esvg in efl recipe and build it "in-tree" between
> eina and evas.

my choices were:

1. esvg and everything it needs come into the efl tree.
2. we are able to build loaders out-of-tree and have them work
3. the esvg svg loader is useless for most and beyond a developer toy may as
well not exist.

unless we make changes beyond just building an efl tree, #3 is what will happen.

-- 
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler)    [email protected]


------------------------------------------------------------------------------
Monitor your physical, virtual and cloud infrastructure from a single
web console. Get in-depth insight into apps, servers, databases, vmware,
SAP, cloud infrastructure, etc. Download 30-day Free Trial.
Pricing starts from $795 for 25 servers or applications!
http://p.sf.net/sfu/zoho_dev2dev_nov
_______________________________________________
enlightenment-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel

Reply via email to