External building (was: Re: [PD] [lanbox] and [tcpclient])
stffn at dibidut.dk
Wed Jan 17 19:51:25 CET 2007
On 17/01/2007, at 18.58, IOhannes m zmoelnig wrote:
> Steffen wrote:
>> On 17/01/2007, at 17.08, IOhannes m zmoelnig wrote:
>>> the best version would be to provide a simple makefile
>> so that people can checkout a single external from cvs plus the pd
>> section (or maybe just pd/src/m_ph.h ?), cd to the dir, run make?
Ok. Dare i ask, what prevents it? I'm guessing that of uniform cross-
platform makefiles. Motivation:
* I just tried to build arraysize. so i checked it out, with pd. But
the makefile shipped in didn't support my current platform plus the
include path wasn't pointing to where it "should". So i had to
changed it, and by inspiration from another makefile (i chose line3)
i got it working. Not too much trouble.
* But passing the right path-var to make that was needed to build
iemmatrix, fx, i don't think i'd worked out my self (i followed
Kevins process due to curiosity on the topic).
To not reinvent sliced bread, how does that idea further differ from
the build-system used by (or maybe more correct: made for) pd-
extended? My guess is, not much, but it currently requires the
packages section of cvs too, if i'm not wrong.
Im not sure where im getting at, except that i think it would be
"nice" if it was "easy" to assemble/get just the settings of Pd one
(the user) wants. There are a few "knobs to adjust" in this domain.
To name some: millers version, pd-extended, devel, dd; pddp, or not;
from none to more or all of the abstractions in cvs; from none to
more or all of the externals in cvs, ...
> if you install pd differently (e.g. on debian with apt), you might get
> the m_pd.h file anyhow, probably as /usr/include/m_pd.h.
> i think no "standard" binary distribution of pd comes without this
> header file. (but i might be wrong).
> so there would be no need to checkout this file for _many_ cases
> will be problems for more in-depth externals that need access to
> g_canvas.h,..; however most externals should be fine with just m_pd.h)
Ahh, i see. Thanks for the insight.
More information about the Pd-list