From mboxrd@z Thu Jan  1 00:00:00 1970
From: rog@vitanuova.com
Message-Id: <200006261652.MAA14372@cse.psu.edu>
To: 9fans@cse.psu.edu
Date: Mon, 26 Jun 2000 17:56:06 +0100
Subject: Re: [9fans] Anybody ported Limbo to Plan9 ?
Topicbox-Message-UUID: cb1eeed8-eac8-11e9-9e20-41e7f4b1d025

> Please tell me also if there is any implementation of .Tk
> library/language on Plan9 ? I looked at Limbo examples of GUI
> programming using .Tk, they are quite nice. It would be great to
> have similar language (may be in form of fileserver, heh) in Plan9.

for programming graphical "widget-based" apps (most plan 9 apps
are not) tk is indeed nice. but it's not available for plan 9 directly.
one thing that makes the use of the inferno tk particularly nice
is the inbuilt support that Limbo has for strings. using a similar
interface from C would make life markedly harder.

> (may be in form of fileserver, heh)

this would actually be possible (would give something of a performance
hit though). if you had a copy of inferno, you could write a limbo app
that served a two-level filesystem containing one directory per top
level window, each containing a cmd file (writes send a tk command;
reads get the return value of the last tk command) and event files, one
per tk channel (reads get incoming events).  something like the above
might work. (given a styx<->9p bridge)

but i'd just use inferno (i would say that, of course!), because i like
limbo and in particular i like the fact that once i've written an app,
it will run without change on plan 9, Windows, Linux or a screen
phone.  having used limbo for a while, it seems so laborious to go back
to C.

> Why I stress on Plan9 more than Inferno, that's because Plan9,
> say so, is more free (in money sense) than Inferno, especially as
> to availability of native OS and its full source.

well, it's true that the source to plan 9 costs less than inferno, but
a) you don't need access to kernel source when developing
under hosted inferno (and all the other source is free)
b) if you've got plan 9 running, then inferno will just work as usual.
c) if you want to port the native kernel to a new device, then i'd suggest
that $300 is not unreasonable (about 4 orders of magnitude cheaper than
it was before...!) compared to all the other incidental costs
involved.

if you get a copy of the CD & manuals we're producing, i think
we're going to include in that a copy of inferno for plan 9.

  cheers,
    rog. (my views, not vita nuova's)