sieve - find in header

[Tom Hendrikx]

On 09-09-16 08:35, Hajo Locke wrote:
> Hello,
> 
> thanks for your help.
> 
> Am 08.09.2016 um 22:13 schrieb Tom Hendrikx:
>> On 08-09-16 20:51, @lbutlr wrote:
>>> On Tue Sep 06 2016 07:25:38 Hajo Locke    <hajo.locke at gmx.de> said
>>>> How to solve this case and find a string in arbitrary/unknown
>>>> headerline?
>>> You’ll have to create multiple sieves covering the possible headers.
>>>
>> Maybe when you tell what you're trying to achieve, we can give you
>> better advice.
>>
>> Almost no one is looking for random strings in random headers, since
>> headers are (somewhat) structured data. Could you share with us what
>> string you are looking and why you don't know up front which header
>> field you need to use to look for it?
> 
> We have a webmailer where experienced users are able to create
> filterrules. They can can choose from a lot of predefined headerlines
> and insert text to find. After that they choose a respective action.
> To cover all cases we had a choosable option "complete header" to find a
> string in every headerline without knowing the specific line. This is
> done easily in procmail:
> 
> :0 H
> * ^*searchstring
> 
> i agree that an option like this is needless and it was a mistake to
> offer it.
> If this is not achievable in sieve, we have to remove this
> search-option. May be we replace by a kind of custom-headerline.

I guess that would be my solution too.

Maybe you can check the existing configurations from your users to see
how many ppl actually use this action. If there any only a few, then
that might be used an an extra argument in the discussion on removing
the option altogether. Especially when you are able to manually replace
the existing uses with a construct that sieve *does* support (f.i.
someone uses this because wasn't able to determine which header name was
actually needed, which is easier when you actually filtered a few messages).

Kind regards,
	Tom