Re: sysinst split project - The Configuration File
Silas Silva <[email protected]>
| Newsgroups | gmane.os.netbsd.devel.installation,gmane.os.netbsd.current |
|---|---|
| Message-ID | <[email protected]> |
On Tue, Dec 15, 2009 at 08:12:45AM +0100, Marc Balmer wrote: > Instead of splitting/rewritting/ sysinstall or replacing it with > bsdinstall, I'd suggest to evolve it: Add Lua for the configuration > and move the logic and user interface from C to Lua, keeping some core > functionality of an installer in C, but make it available to Lua. A > good installer would also let me install packages, maybe even meta- > packages, so that one of the first questions this installer asks, is: > "what kind of system do yo want to install?" and provides me with a > selection of meta-packages that are available on the distribution > medium (and for which the dependent packages are also available for > the architecture given). > > Then the ideal installer would allow interactive operation or complete > unattended operation, where installation instructions are give in a > file. > > Well, the ideal installer does not yet exist, but ahoka and myself > have some loose plans to make exactly this happen... Hey Marc and all. Well, I was in doubt after reading the Turing-completeness problem of configuration vs. scripting and whether or not sysinst.conf should be Turing complete. But I mulled over it and I think using Lua on the project would be a good thing. One of the advantages of it is that we could trust the Lua parser, so it would not be necessary to create one more parser in Lex & Yacc as I previously suggested (although I'd love to do that :-P). Bringing Lua wields a bigger question: shouldn't NetBSD import to its tree, a powerful general-purpose interpreter? (sh, ksh, awk, bc, etc., are not general-purpose). Why not just import Lua to the base system so people can built projects on it, like sysinst? Just an idea. I've already used Lua embedded in a C++ program, not too much, but in a level enough to note that it works great and glue very well. Your ideas for sysinst are great. I would love to help on this project! Well, I'm not going to repost all the XML pro and cons that people posted here, but I disagree with the adoption of XML. However, since I'm much worried about helping bringing a new sysinst to live, I'll work on any solution, whatsoever. Couldn't sysinst import/export from/to both formats, human-readable (Lua script or another) and XML? That would solve the controversy, but it may be difficult to maintain support for two different standards. Thank you very much! -- Silas Silva