Re: min() max() math functions (was: Re: Feature request (@M):# with context matches)

Re: min() max() math functions (was: Re: Feature request (@M):# with context matches)

[Bart Schaefer]

[>workers]

On Feb 9,  3:23am, Daniel Shahaf wrote:
}
} I said "built-in" but I meant in zsh/mathfunc, with all other math
} functions, for consistency.

I was talking about zsh/mathfunc, too.

} I suppose I could package them as an autoload.  Patch below. 

That's what I was going to suggest, even to the creation of Functions/Math/.

} How do I shield zmathfunc against callers that unsetopt functionargzero?

Sticky emulation would do it for you.  E.g.

    emulate zsh -c 'autoload -U zmathfunc'

Or examine $funcstack[1] instead of $0 (with localoptions noksharrays),
although that introduces a dependence on zsh/parameter.

} I guess I'll have to drop the the multifuncdef usage altogether?

See for example zmv, zcp, zln.

Although it seems an oversight that you can't combine -u and -M for the
functions command.

} +Autoloading tt(zmathfunc) defines the three mathematical functions

You do have to both autoload it and run it, like e.g. "colors".  Given
that, why not have zmathfunc perform "zmodload zsh/mathfunc" too?

} +++ b/Doc/Zsh/mod_mathfunc.yo
} +The functions tt(min), tt(max), and tt(sum) are defined not in this module
} +but in the tt(zmathfunc) autoloadable function

Ehh, I'm not sure it's the job of any module's doc to explain what it
does NOT define, much less to cross-reference zshcontrib.

Re: min() max() math functions (was: Re: Feature request (@M):# with context matches)

[Daniel Shahaf]

Bart Schaefer wrote on Mon, Feb 08, 2016 at 21:22:22 -0800:
> On Feb 9,  3:23am, Daniel Shahaf wrote:
> } How do I shield zmathfunc against callers that unsetopt functionargzero?
> 
> Sticky emulation would do it for you.  E.g.
> 
>     emulate zsh -c 'autoload -U zmathfunc'
> 

Good to know.

If I'm not mistaken, using this form would also fix the cshjunkiequotes
problems Sebastian has been running into.  

Shouldn't this form be used more often?  As in, wouldn't using this form
be correct for the majority of autoloadable functions?

> Or examine $funcstack[1] instead of $0 (with localoptions noksharrays),
> although that introduces a dependence on zsh/parameter.
> 
> } I guess I'll have to drop the the multifuncdef usage altogether?
> 
> See for example zmv, zcp, zln.
> 

I think the path of least resistance is to drop the multifuncdef usage,
and not require the user to create symlinks or to have zsh/parameter
[although that module is probably ubiquitous?] or to do the autoload
specially.

> Although it seems an oversight that you can't combine -u and -M for the
> functions command.
> 

Would this be easy to add?  (Just another flag in the stub function
structure, maybe?)  I can check later but don't have time to right now.

> } +Autoloading tt(zmathfunc) defines the three mathematical functions
> 
> You do have to both autoload it and run it, like e.g. "colors".

Indeed.  How about:

    The function tt(zmathfunc) defines the three mathematical functions...

> Given that, why not have zmathfunc perform "zmodload zsh/mathfunc"
> too?
> 

I don't see a good reason to force users of the autoload to use the
module.  It's a non-obvious side effect (consider somebody who does
'autoload zmathfunc && zmathfunc && zmodload -F ... zsh/mathfunc')
that's not easy for the caller to disable.  (But see below)

> } +++ b/Doc/Zsh/mod_mathfunc.yo
> } +The functions tt(min), tt(max), and tt(sum) are defined not in this module
> } +but in the tt(zmathfunc) autoloadable function
> 
> Ehh, I'm not sure it's the job of any module's doc to explain what it
> does NOT define, much less to cross-reference zshcontrib.

The functionality of zmathfunc is related to that of zsh/mathfunc.  The
reason they aren't the same thing is due to implementation considerations
(we separated functions that can be implemented satisfactorily in shell
code from those that must be implemented in C), not due to interface
considerations.  In fact, that's an argument in favour of making
zsh/mathfunc autoload zmathfunc (sic, in this order).

Why is it a problem for zshmodules to reference zshcontrib?  Should
zmathfunc's documentation be located elsewhere?

Cheers,

Daniel

Re: min() max() math functions (was: Re: Feature request (@M):# with context matches)

[Bart Schaefer]

On Feb 10,  9:23am, Daniel Shahaf wrote:
} Subject: Re: min() max() math functions (was: Re: Feature request (@M):# w
}
} Bart Schaefer wrote on Mon, Feb 08, 2016 at 21:22:22 -0800:
} > 
} >     emulate zsh -c 'autoload -U zmathfunc'
} 
} Shouldn't this form be used more often?  As in, wouldn't using this form
} be correct for the majority of autoloadable functions?

It has other side-effects which are desirable in the zmathfunc case but
may not be in other cases.  (Specifically, the stickyness is transitive
to functions defined when the autoloaded function runs.)

Whether "the majority" of autoloads would benefit, I don't know.  Some
functions perform a task in a very isolated way, others are intended
to preserve the behavior the user's setopts have defined, and it's not
possible to readily determine which have the latter intent.

} > Although it seems an oversight that you can't combine -u and -M for the
} > functions command.
} 
} Would this be easy to add?

I took a quick look and decided it wasn't straightforward enough to mess
with today.  The different interpretation of the argument list for -M
makes it complicated to untangle the current implementation.  And one
can get the same effect by doing 

    autoload mathfunc
    functions -M mf 0 0 mathfunc

in either order (that is, the shell function doesn't have to be defined
to link it to a math context name).

} (Just another flag in the stub function structure, maybe?)

Doesn't even need that.

} > Ehh, I'm not sure it's the job of any module's doc to explain what it
} > does NOT define, much less to cross-reference zshcontrib.
} 
} Why is it a problem for zshmodules to reference zshcontrib?  Should
} zmathfunc's documentation be located elsewhere?

It's not exactly a problem, more like a bad precedent.  E.g. I was careful
to have zshcontrib for the *-word-match functions refer to the vim text
objects, but I would not reference in the other direction.  It should IMO
be possible to throw away the zshcontrib manual and still have a complete
reference to the "bare" shell.

The most obvious counter-argument is that the entire completion system
depends on mutually cooperating options, builtins, and shell functions.
And the complist module already references the colors autoload, so I
should probably just shrug and ignore my sense of propriety.

Re: min() max() math functions (was: Re: Feature request (@M):# with context matches)

[Daniel Shahaf]

Bart Schaefer wrote on Wed, Feb 10, 2016 at 09:41:34 -0800:
> On Feb 10,  9:23am, Daniel Shahaf wrote:
> } > Ehh, I'm not sure it's the job of any module's doc to explain what it
> } > does NOT define, much less to cross-reference zshcontrib.
> } 
> } Why is it a problem for zshmodules to reference zshcontrib?  Should
> } zmathfunc's documentation be located elsewhere?
> 
> It's not exactly a problem, more like a bad precedent.  E.g. I was careful
> to have zshcontrib for the *-word-match functions refer to the vim text
> objects, but I would not reference in the other direction.  It should IMO
> be possible to throw away the zshcontrib manual and still have a complete
> reference to the "bare" shell.
> 
> The most obvious counter-argument is that the entire completion system
> depends on mutually cooperating options, builtins, and shell functions.
> And the complist module already references the colors autoload, so I
> should probably just shrug and ignore my sense of propriety.

I don't think it's a problem for other manuals to provide pointers to
functionality in zshcontrib.  It might be a problem if some core
functionality _depended_ on zshcontrib functionality, but that's not the
case here.

In the interest of going forward, I'll commit what I have so far, but we
can tweak the docs in further commits if needed.

Cheers,

Daniel




Interdiff relative to users/21256:

diff --git a/Doc/Zsh/contrib.yo b/Doc/Zsh/contrib.yo
index 11678e5..f8754fb 100644
--- a/Doc/Zsh/contrib.yo
+++ b/Doc/Zsh/contrib.yo
@@ -3804,7 +3804,7 @@ xitem(tt(min+LPAR())var(arg)tt(, ...+RPAR()))
 xitem(tt(max+LPAR())var(arg)tt(, ...+RPAR()))
 xitem(tt(sum+LPAR())var(arg)tt(, ...+RPAR()))
 item(tt(zmathfunc))(
-Autoloading tt(zmathfunc) defines the three mathematical functions
+The function tt(zmathfunc) defines the three mathematical functions
 tt(min), tt(max), and tt(sum).  The functions tt(min) and tt(max) take
 one or more arguments.  The function tt(sum) takes zero or more arguments.
 Arguments can be of different types (ints and floats).
diff --git a/Functions/Math/zmathfunc b/Functions/Math/zmathfunc
index 17d3336..4ff4070 100644
--- a/Functions/Math/zmathfunc
+++ b/Functions/Math/zmathfunc
@@ -1,21 +1,26 @@
 #autoload
 
-setopt localoptions multifuncdef
+zsh_math_func_min() {
+  local result=$1
+  shift
+  local arg
+  for arg ; do
+    (( $arg < result )) && result=$arg
+  done
+  (( result )) # return
+}
+functions -M min 1 -1 zsh_math_func_min # at least one argument
 
-zsh_math_func_min zsh_math_func_max() {
+zsh_math_func_max() {
   local result=$1
   shift
   local arg
   for arg ; do
-    case $0 in
-      (*max) (( $arg > result )) && result=$arg;;
-      (*min) (( $arg < result )) && result=$arg;;
-    esac
+    (( $arg > result )) && result=$arg
   done
   (( result )) # return
 }
 functions -M max 1 -1 zsh_math_func_max # at least one argument
-functions -M min 1 -1 zsh_math_func_min # at least one argument
 
 zsh_math_func_sum() {
   local sum