Re: (wrong-type-argument number-or-marker-p nil) error at gnus startup

[Hirotaka Honda <hirotaka@hondalaw.com>]

Dear all

After I commented out gnus-registry as the following, any errors (or the same errors) happened again.

;; (gnus-registry-initialize)
;; (setq
;;  gnus-registry-split-strategy 'majority
;;  gnus-registry-ignored-groups '(
;; 								("nntp" t)
;; 								("nnrss" t)
;; 								("mail.spam" t)
;; 								("archive" t)
;; 								("Send-Mail" t))
;; gnus-registry-track-extra nil
;; gnus-registry-max-entries 500000)

;; (setq spam-log-to-registry t)

Hirotaka Honda

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> On 06/01/21 22:30 PM, Hirotaka Honda wrote:
>> Dear all
>>
>> I am not certian what is happening, though the most dubious mail has
>> the email header like this:
>>
>> X-From-Line: - Fri May 28 HH:MM:SS 2021 [I masked the time]
>> X-Mozilla-Status: 0001
>> X-Mozilla-Status2: 00000000
>> ...
>> From: =?iso-2022-jp?B?[mask]?= <[mask]@outlook.jp>
>> To: =?utf-8?B?[mask]?=
>> =?utf-8?B?[mask]?= <[mask]@[mask]>
>> Cc: =?iso-2022-jp?B?[mask]?=
>> =?iso-2022-jp?B?[mask]=?= <[mask]@[mask]>,
>> "[mask]@[mask]" <[mask]@[mask]>
>> ...
>> X-Mailer: JavaMail
>>
>> Does the Gnus expect that To field and Cc field are coded in different ways?
>
> That shouldn't be a problem, the decoding process is called once for
> each header, and each time it detects the coding over again.
>
> My only uncertainty is, should the headers somehow be already decoded at
> this point (during registry operation), and if so and we're just missing
> a decode call, why hasn't this become an issue in the past? I suspect we
> just need to add decoding to the process, but it still makes me suspicious.