put value last in *Help* buffer for C-h v?

put value last in *Help* buffer for C-h v?

[Drew Adams]

I wonder if it wouldn't be better to systematically put the variable's value
last in the *Help* buffer for `C-h v'?

Consider, for example, what you see for something like
`font-lock-syntax-table':

 font-lock-syntax-table is a variable defined in `font-lock.el'.
 Its value is shown below.

 Documentation:
 Non-nil means use this syntax table for fontifying.
 If this is nil, the major mode's syntax table is used.
 This is normally set via `font-lock-defaults'.

 Value:
 #^[t nil
      (3)
      (3)
 ...
         (2)
         nil syntax-table]
      syntax-table]

 Local in buffer font-lock.el; global value is nil

 [back]

The "..." here represents 12,228 lines. This is an extreme case, of course,
but wouldn't it always be best to put the value last? The information about
the local and global values, and the [back] button, are more convenient if
presented before the value, no?

We might also consider placing a second [back] button after the value, if it
is longer than, say, 50 lines.

Re: put value last in *Help* buffer for C-h v?

[Mathias Dahl]

> I wonder if it wouldn't be better to systematically put the variable's value
> last in the *Help* buffer for `C-h v'?

I think I agree. It is a bit strange to see the value at the beginning
of the buffer sometimes and sometimes not (because it is "large").
Shouldn't we strive for a consistent user interface if we can?

> but wouldn't it always be best to put the value last? The information about
> the local and global values, and the [back] button, are more convenient if
> presented before the value, no?

I agree, it seems logical to keep all information about the variable
together, in one block.

> We might also consider placing a second [back] button after the value, if it
> is longer than, say, 50 lines.

I have no strong opinion on this apart from thinking that this too is
inconsistent; as a user I want to be able to trust that the same user
interface widgets are available each time I execute a certain
operation. However, I it might look goofy to see something like this:

  [back]
  Value:
   t
  [back]

/Mathias

RE: put value last in *Help* buffer for C-h v?

[Drew Adams]

> > We might also consider placing a second [back] button after the
> > value, if it is longer than, say, 50 lines.
>
> I have no strong opinion on this apart from thinking that this too is
> inconsistent; as a user I want to be able to trust that the same user
> interface widgets are available each time I execute a certain
> operation. However, I it might look goofy to see something like this:
>
>   [back]
>   Value:
>    t
>   [back]

That's why I said "if it is longer than, say, 50 lines." In the case you
gave, the value takes up only one line.

Make it 1000 lines instead of 50, if you like. I figured that if you have to
scroll down, then another [back] button could be helpful.

Re: put value last in *Help* buffer for C-h v?

[Richard Stallman]

    I think I agree. It is a bit strange to see the value at the beginning
    of the buffer sometimes and sometimes not (because it is "large").
    Shouldn't we strive for a consistent user interface if we can?

Consistency is not the only criterion.  When the value is short,
it is convenient to see it near the top, so that you don't have
to scroll down to find it.

Re: put value last in *Help* buffer for C-h v?

[Mathias Dahl]

> >   [back]
> >   Value:
> >    t
> >   [back]
>
> That's why I said "if it is longer than, say, 50 lines." In the case you
> gave, the value takes up only one line.

Sorry for being unclear; my point was that we SHOULD, in principle,
always print out the second [back], because of consistency. But at the
same time I do not agree with myself because I think my own "rule"
would create goofy situation like above. In other words, don't listen
to me...

> Make it 1000 lines instead of 50, if you like. I figured that if you have to
> scroll down, then another [back] button could be helpful.

Yes, and I agree.

Re: put value last in *Help* buffer for C-h v?

[Mathias Dahl]

> Consistency is not the only criterion.  When the value is short,
> it is convenient to see it near the top, so that you don't have
> to scroll down to find it.

True, it is convenient if the description is so large that the value
will end up below so that you have to scroll.