Re: Why asterisks in nnmail-split-fancy rule?

Re: Why asterisks in nnmail-split-fancy rule?

[Svend Tollak Munkejord]

Today, Kai Großjohann <kai.grossjohann@gmx.net> wrote:

> Svend Tollak Munkejord <stm@bacchus.pvv.org> writes:
>
>> What is the difference, in nnmail-split-fancy, between
>>
>> (from "foo" "mail.foo")
>>
>> and
>>
>> (from ".*foo.*" "mail.foo") ?
>
> Fancy splitting implicitly adds \\<...\\> around the value.  This
> means that you can't use "xy@" as the value: because "@" is not a
> word constituent, a word can never end there, so "xy@\\>" will never
> match.  That's why you have to use "xy@.*".

OK, but I did not use any "@", only a surname string. I am puzzled, because
the first rule (without .*) worked on a message from "Peter Foo", but not
on a message from "Mike Foo", and by staring at the headers of those
messages, I cannot figure out where this effect comes from.

-- 
Svend Tollak Munkejord 

Re: Why asterisks in nnmail-split-fancy rule?

[Svend Tollak Munkejord]

I wrote:

> Today, Kai Großjohann <kai.grossjohann@gmx.net> wrote:
>
>> Svend Tollak Munkejord <stm@bacchus.pvv.org> writes:
>>
>>> What is the difference, in nnmail-split-fancy, between
>>>
>>> (from "foo" "mail.foo")
>>>
>>> and
>>>
>>> (from ".*foo.*" "mail.foo") ?
>>
>> Fancy splitting implicitly adds \\<...\\> around the value.  This
>> means that you can't use "xy@" as the value: because "@" is not a
>> word constituent, a word can never end there, so "xy@\\>" will never
>> match.  That's why you have to use "xy@.*".
>
> OK, but I did not use any "@", only a surname string. I am puzzled,
> because the first rule (without .*) worked on a message from "Peter Foo",
> but not on a message from "Mike Foo", and by staring at the headers of
> those messages, I cannot figure out where this effect comes from.

And I forgot to say that the strange message also really needed the first
".*" in ".*foo.*".

-- 
Svend Tollak Munkejord 

Re: Why asterisks in nnmail-split-fancy rule?

[Kai Großjohann]

Svend Tollak Munkejord <stm@bacchus.pvv.org> writes:

> From: =?iso-8859-1?q?Mike=20Foo?= <someaddress@yahoo.no>
>
> (This is a bit strange, since the name does not any øæå or other `special'
> characters).
>
> Is this the reason why I would need the ".*"s? I do not understand this,
> since the string "foo" is still there.

In the string "20Foo", F is not the beginning of a word, so the
string does not match "\\<Foo\\>".

-- 
This line is not blank.

Re: Why asterisks in nnmail-split-fancy rule?

[Svend Tollak Munkejord]

On 2003-05-30, Kai Großjohann <kai.grossjohann@gmx.net> wrote:

> Svend Tollak Munkejord <stm@bacchus.pvv.org> writes:
>
>> From: =?iso-8859-1?q?Mike=20Foo?= <someaddress@yahoo.no>
>>
>> (This is a bit strange, since the name does not any øæå or other
>> `special' characters).
>>
>> Is this the reason why I would need the ".*"s? I do not understand this,
>> since the string "foo" is still there.
>
> In the string "20Foo", F is not the beginning of a word, so the
> string does not match "\\<Foo\\>".

Ah, I see. Thanks for your help!

-- 
Svend Tollak Munkejord 

Re: Why asterisks in nnmail-split-fancy rule?

[Kai Großjohann]

Josh Huber <huber@alum.wpi.edu> writes:

> It seems like many people are confused by this...what is the rationale
> behind it?  Was the purpose to make it not-so-surprising to people who
> don't know regular expressions?

It was an attempt at DWIM.  But DWIM is difficult.

With this, you could put (from "josh" "bla") and (from "joshua"
"lala") into nnmail-split-fancy and not be surprised.

Of course, (from "josh" "bla") and (from "josh-smith" "lala") will
still be surprising.

However, I find that it works sufficiently well to just use simple
words in the rules, it does what I need it seems.
-- 
This line is not blank.

Re: Why asterisks in nnmail-split-fancy rule?

[Michael R. Wolf]

kai.grossjohann@gmx.net (Kai Großjohann) writes:

> Michael R. Wolf <MichaelRunningWolf@att.net> writes:
>
>> I understand. That's how I found it. In addition, I think that
>> restating the need for restriction at the message-fetch-field level
>> would be helpful.
>
> Documented.  Will commit soon.

Thanks, but as Benjamin suggested, I'd prefer to see his *total*
solution get applied. He did a great job of researching where the code
changes should be made to bring it into alignment with the doc change.
And he created a patch, too.

-- 
Michael R. Wolf
    All mammals learn by playing!
        MichaelRunningWolf@att.net

Re: Why asterisks in nnmail-split-fancy rule?

[Kai Großjohann]

Benjamin Rutt <rutt+news@cis.ohio-state.edu> writes:

> While we're on the topic, could you instead consider the patch I sent
> recently to the ding list?  My patch fixes the doc for
> `message-fetch-field' as well as fixes a number of bugs in usages of
> `message-fetch-field'.
>
> It's a little frustrating to see my patch ignored, and then see part
> of my efforts reproduced a week later.

*blush*

I continually hit M-u on that message to keep it on my todo list.
Maybe I should take the chance on the pentecost weekend to take a
break from searching for a new job and relax a bit instead with Gnus
hacking and patch integration and debugging and stuff...  (I still
wonder where that memory is going when I `J s'...)
-- 
This line is not blank.