On Apr 8, 2009, at 3:23 PM, Frank Barknecht wrote:
Hallo, Hans-Christoph Steiner hat gesagt: // Hans-Christoph Steiner wrote:
On Apr 8, 2009, at 5:35 AM, cyrille henry wrote:
i think for this specific lib, any dependency of object outside it's directory can be seen as a bug.
I don't understand this at all. Why did every modern operating
system spend massive amounts of work to switch from static linking of
libraries (i.e. including the library with the program itself) to dynamic,
shared libraries (i.e. a single copy of a library shared by all programs
that use it)?I think, advantages and disadvantages have to be weighted, and IMO
mapping currently is leaning to a not very user friendly side. For example
its use of [float_argument], [symbol_argument] and [once] from purepd creates a
dependency on purepd that is completely unnecessary: [once] is only used once, [symbol_argument] also, and float_argument very often is used where
a simple [f $1] would be enough (its second argument is empty). So I'd say, get
rid of it and be more friendly to users.
If there is good code out there, I want to use it, not reinvent it.
If you use mapping in Pd-extended, you never need to know anything
about what's happening inside, it just works. With a little bit of
effort, Pd-vanilla users can also acheive this. That's why I am
fixing up libdir.c and these libraries in SVN.
Another case is the "standardization" to use abstraction names, that
require the (sometimes broken) hexloader: Life would be much easier for a
lot of people without the "->" convention.
Indeed, I fully agree. That's why I fixed it over a year ago :D
mapping needs to hexloader tricks.
Other cases may not be so easy to fix and some externals are hard to
replace, but as it currenly is, mapping looks like a collection that puts
some abstract design principles higher than actual user demands. Of course that's
just my humble opinion, which is leaning to a much more simplistic side.
These libraries were a place for me to experiment with ways of
handling libraries in Pd easier. Some of the experiments failed, but
I think right now its in a pretty good place. So try it out the way
it is now and tell me what doesn't work.
.hc
Ciao
Frank
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
All information should be free. - the hacker ethic