Date: Wed, 17 Apr 2002 10:20:13 -0400 From: Antoine Beaupre <anarcat@anarcat.ath.cx> To: Max Okumoto <okumoto@ucsd.edu> Cc: The Anarcat <anarcat@anarcat.dyndns.org>, freebsd-libh@FreeBSD.ORG Subject: Re: cleanup of HSystem 2nd try (stage 4) Message-ID: <3BABB735-520E-11D6-BCD9-0050E4A0BB3F@anarcat.ath.cx> In-Reply-To: <hfy9fm4eez.fsf@multivac.sdsc.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
Le Mercredi 17 avril 2002, =E0 09:56 , Max Okumoto a =E9crit : > The Anarcat <anarcat@anarcat.dyndns.org> writes: > > [stuff deleted] >> I don't know if anyone ever considered replacing the current tcl >> interface stuff with swig, but it's an interesting idea. I never >> really paid attention to swig before. I think, however, that reducing >> the number of dependencies of libh would help it getting into the = base >> system. There's already a big issue about libh using tcl as a >> scripting language because tcl itself is not part of base, and >> everyone seems to hate tcl. :) > > I don't have many problems with tcl. Quoting can become a pain > but otherwise its just another programming language. That's what I think too. > Swig (http://www.swig.org/) is a software development tool > that connects programs written in C, C++, and Objective-C > with a variety of high-level programming languages. (Java, > Perl, Tcl/Tk, etc) Yeah, I read their propaganda too. The advantage of swig over our current scheme is that it kind of removes=20= the binding of libh with tcl. It allows generation of other interfaces.=20= But that can also be a problem in itself: tcl has the safe interpreter=20= mode for installing packages which is mostly why it was chosen (along=20 with easy C integration). But then again, if the packages are written in=20= TCL, they won't be interpreted as perl, won't they? :) The disadvantage of swig over our current scheme is that it's another=20 dependency, and we don't have control over the API. > After I have clean HSystem up. I might consider replacing it with > swig. But first I have to understand it first. :-) Yeah. And you'll have to face the flamewar. But I think I'd be happy to=20= see a flamewar on libh@. It'd warm up the project. ;) >> Did you mention you have a non-perl parser that replace >> find_classes_descriptions.pl? That would sure be interesting. Again, >> the more we reduce dependencies, the more libh will get >> acceptance. But perl isn't really considered as excessive dependency >> since it's in the base system. > [stuff deleted] > > Yea, the parser would be in HSystem. So we could get ride of a few > steps in the build process. Basically build_systems_*.cc and > tcl_interface_gen_* would go away. Whoa, that would be good. Really good. It's really bad to have generated=20= cc files in the compile directories: they escape documentation and=20 impede understanding of the system. > *.cd.cc -> build_systems_*.cc -> tcl_interface_gen_* ->=20 > LibTclInterface_*.cc > would become > > *.cd.cc -> LibTclInterface_*.cc I don't see anything wrong with this, if it works. I was wondering why=20= the heck we'd parse the .cd.cc files with a perl script to generate C++=20= code when the .cd.cc file is C++ code in itself. > Max Okumoto , my hero. ;) A. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-libh" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3BABB735-520E-11D6-BCD9-0050E4A0BB3F>