Reply to message «Re: Vim Script grammer», sent 00:27:41 12 July 2011, Tuesday by Marc Weber:
> @ ZyX: compile it? hell. compile to JS and use a JS language. JS is > known by everyone today! LLVM has JS backend. LLVM enables you to write both compiler and interpreter in one turn (LLVM-jit technology is intended to build just the same program as JS compilers in browser). The reasons why VimL-LLVM-jit should be written is 1. I want to once learn how to create compilers. 2. Current VimL implementation is slow. 3. VimL already has large codebase of editor tricks. Of course this will require also creating a new editor. > VimL is simple so you can get it done. But you seriously should have a > look at all those other languages such as F# first. > They should be as productive as VimL if you think about it being a DSL. > And they are way superior if you want to get more things done. They don't have that large codebase of useful editor features. You once pointed me to Yi, why you are not using it as much as vim? I guess answer is written above. Original message: > @ ZyX: compile it? hell. compile to JS and use a JS language. JS is > known by everyone today! > Don't reinvent the wheel VimL is no longer what it should be... :( > We should push an existing language. We don't want to force the world to > learn yet another DSL for configuring editors only. > > If you want to optimize something: Add a regex native type so that regex > don't have to be recompiled on each comparison, please. This would speed > up quite a lot. > > Why JavaScript? Because many advanced implementations already exist. > Because most languages can be compiled to JS (because everyone wants to > target browsers anyway) ... > > (My 2 cents on this) > > Excerpts from Charles Strahan's message of Mon Jul 11 20:27:17 +0200 2011: > > Ultimately, I want ViEmu (or an alternative Visual Studio plugin) to have > > proper (relatively complete) support for VimL scripts. > > Hell. Tell me why. There are many nice languages available for .net > which are ways superior to VimL (which was ahead of its time 30 years > ago..) .. > > Also think about whether you only want VimL or VimL and Vim's api (:h > functions). If you want all of them .. good luck. I don't think this > makes much sense except functions like getline or getcursor. > > Most features will be impossible (or very hard) to implement unless > you're going to spend months on it. > > Be aware about VimL's exec command. Very often the VimL API is poorly > designed. Eg you have to do: > > exec 'set rtp+='.escape(path,',') > > or the like. it should have been: > > call SetRTPath([path]++oldpath) > > You don't want to implement all this (sh..) > > > I'm sure it would be a ton of work, but it looks like good fun. > > VimL is simple so you can get it done. But you seriously should have a > look at all those other languages such as F# first. > They should be as productive as VimL if you think about it being a DSL. > And they are way superior if you want to get more things done. > > Of course - that's only my point of view :) You may have your own valid > differing judgments. > > Marc Weber
signature.asc
Description: This is a digitally signed message part.
