[Toybox] awk (Re: ps down, top to go)

[Roy Tam]

2016-05-31 12:36 GMT+08:00 Andy Chu <andychup at gmail.com>:
> (+toybox list again, since I think you intended that)
>
> On Mon, May 30, 2016 at 8:26 PM, Roy Tam <roytam at gmail.com> wrote:
>> For Coherent 4.2 /bin/sh, please check this out:
>> https://github.com/roytam1/mwc-sh
>> Since the github mirror doesn't extract everything so I have to do it myself.
>
>> May be you missed yash shell.
>> https://yash.osdn.jp/index.html.en
>
>
> OK, that didn't take long!  Thanks Roy.

and the latest minix /bin/sh is here:
https://github.com/minix3/minix/tree/R3.1.6/commands/sh

>
> I looked at both of these, and they are indeed quite complete POSIX
> compliant Bourne shells.  Coherent sh is ~11K LOC, while yash is ~43K
> LOC, including its own line editing library.
>
> Coherent sh actually uses yacc like bash, and I believe this comment
> in parse.y is related to the same pitfalls that bash ran into:
>
> """
> In the original grammar, no distinction between simple command and
> compound
> commands was made. This, along with the right-recursive formulation of
> the
> command grammar, created a need for lookahead that defeated the
> complex
> machinery for context-sensitive lexing that is required.
> """
>
> My reading of this is: "we subsetted the POSIX grammar, because if you
> use the full grammar, it becomes really difficult to parse command
> substitution -- matching $( and )".  bash solved this by duplicating a
> lot of the parser, as previously discussed.
>
> In my defense, Coherent sh was only open sourced in 2015!  And
> development apparently stopped in 1995.
>
> yash looks like it started as a student project in 2007, and has
> evolved into a comprehensive POSIX compliant shell, and is still under
> active development.  It even has an Ubuntu package (yash)!
>
> I updated this section of Wikipedia based on this new information:
> https://en.wikipedia.org/wiki/Unix_shell#Bourne_shell
>
> (I had done a lot of research on this, including in some books, but
> yeah I definitely missed those two completely.)
>
> OK, so does anyone want to dispute my new claim that there are only
> 6-7 open source code lineages, and 3-4 started without paid labor
> (hush is still TBD)?
>
> Andy