`W w' (gnus-article-fill-cited-article)

`W w' (gnus-article-fill-cited-article)

[David Abrahams]

I've been noticing some inconvenient behavior from the wrap-washing
function of the subject line.  I applied it to a message containing:

  >> Like you Peter, I've been trying to explore the use cases of
  >> move_ptr/scoped_ptr.  In my own code I use these types of smart
  >> pointers only in a scope where under non-exceptional conditions the
  >> smart pointer is moved from (release'd) to client code.  Only under
  >> exceptional conditions does the scoped_ptr actually delete anything.
  >> In this context, client code must know exactly how the resource is
  >> allocated and deallocated.
  >
  > True enough, and the primary use of auto_ptr/move_ptr is to return
  > things
  > from a factory function when the client needs to know exactly how the
  > resource has been allocated. That's why I think that an enhanced
  > scoped/move/auto_ptr needs its deallocator encoded in the type as a
  > template
  > parameter, so that the client knows what resource management scheme
  > they are
  > dealing with.


And what I got out was:


  >> Like you Peter, I've been trying to explore the use cases of
  >> move_ptr/scoped_ptr.  In my own code I use these types of smart
  >> pointers only in a scope where under non-exceptional conditions the
  >> smart pointer is moved from (release'd) to client code.  Only under
  >> exceptional conditions does the scoped_ptr actually delete
  >> anything.  In this context, client code must know exactly how the
  >> resource is allocated and deallocated.
  > True enough, and the primary use of auto_ptr/move_ptr is to return
  > things from a factory function when the client needs to know exactly
  > how the resource has been allocated. That's why I think that an
  > enhanced scoped/move/auto_ptr needs its deallocator encoded in the
  > type as a template parameter, so that the client knows what resource
  > management scheme they are dealing with.

In other words, the blank line between the two separate quotations
disappeared.  Is there a way to prevent this from happening?

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com

Re: `W w' (gnus-article-fill-cited-article)

[Lars Magne Ingebrigtsen]

David Abrahams <dave@boost-consulting.com> writes:

> In other words, the blank line between the two separate quotations
> disappeared.  Is there a way to prevent this from happening?

It's apparently something that `fill-region' does.  `fill-prefix' is
bound (by Gnus) to be (in this instance) "> ", and `fill-region'
seemingly wipes out leading lines that consist only of the prefix...

I think the proper fix is probably something in the filling code, but
Gnus could cheat by not including the leading line in the region to
be filled.  The consequences might not always be what we want,
though... 

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: `W w' (gnus-article-fill-cited-article)

[David Abrahams]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> David Abrahams <dave@boost-consulting.com> writes:
>
>> In other words, the blank line between the two separate quotations
>> disappeared.  Is there a way to prevent this from happening?
>
> It's apparently something that `fill-region' does.  `fill-prefix' is
> bound (by Gnus) to be (in this instance) "> ", and `fill-region'
> seemingly wipes out leading lines that consist only of the prefix...
>
> I think the proper fix is probably something in the filling code, but
> Gnus could cheat by not including the leading line in the region to
> be filled.  

I'm (just guessing) but does Gnus try to fill the entire message as
one region? It probably ought to be finding individual paragraphs to
avoid this problem.  I also find some issues in filling where people
fail to leave a blank line before replies:
   
   > Somebody writes:
   > > something or other
   > and I reply with something else

Actually, I would be very happy if washing would add those blank
lines so that I can see the separate parts of the thread:

   > Somebody writes:
   >
   > > something or other
   >
   > and I reply with something else

> The consequences might not always be what we want,
> though... 

Thanks very much for thinking about this; I'll be looking forward to a
resolution.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com

Re: `W w' (gnus-article-fill-cited-article)

[Lars Magne Ingebrigtsen]

David Abrahams <dave@boost-consulting.com> writes:

> I'm (just guessing) but does Gnus try to fill the entire message as
> one region? It probably ought to be finding individual paragraphs to
> avoid this problem.

No, it narrows to regions (as determined by gnus-cite), and does
filling on one region at a time.

> Actually, I would be very happy if washing would add those blank
> lines so that I can see the separate parts of the thread:

I think that sounds like new functionality -- adding a blank line
between each cited part.  It shouldn't be too difficult to do if
anybody wants to give it a whirl -- just loop over
`(gnus-dissect-cited-text)' and insert some blank lines.  See, for
instance, `gnus-article-fill-cited-article' as an example.

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen