Should gnus-summary-repair-multipart modify files on disk? (was: New summary part commands?)

[Reiner Steib <4.uce.03.r.s@nurfuerspam.de>]

On Wed, Feb 19 2003, Kai Großjohann wrote:

> Reiner Steib <4.uce.03.r.s@nurfuerspam.de> writes:
>> BTW, for `gnus-summary-repair-multipart' I think we should add a
>> `y-or-no-p' because it may *destroy* an article (on writable back
>> ends), e.g. if the original article wasn't b0rked.
>
> Really?  I was shocked, so I quickly scanned the code, 

,----[ C-h f gnus-with-article RET ]
| gnus-with-article is a Lisp macro in `gnus-sum'.
| (gnus-with-article ARTICLE &rest FORMS)
| 
| Select ARTICLE and perform FORMS in the original article buffer.
| Then replace the article with the result.
`----

Note the last sentence.

> but it appears to only modify the buffer, not the file on disk.

It *does* modify the article disk if you use nnfolder (Article
<news:3D2BFC8C.BeroList-2.5.9@wpxx02.toxi.uni-wuerzburg.de> from
Gmane) copied to nnfolder:Test and `K m'-ed:

--8<---------------cut here---------------start------------->8---
$ diff -U1 multi?.mbox
--- multi1.mbox Wed Feb 19 21:06:34 2003
+++ multi2.mbox Wed Feb 19 21:06:47 2003
[...]
@@ -12,4 +12,2 @@
[...]
-Mime-Version: 1.0
-Content-Type: multipart/signed; micalg=pgp-sha1;
[...]
@@ -40,4 +38,8 @@
[...]
+Mime-Version: 1.0
+Content-Type: multipart/mixed; boundary="mP3DRpeJDSE+ciuQ"
[...]
--8<---------------cut here---------------end--------------->8---

In this case, the repairing was successful.  But after reading
<news:m3bs1jppi0.fsf@quimbies.gnus.org>, I did some tests and at least
once, the `repaired' article was corrupt.  Maybe I'm able to reproduce
it later.  I didn't think that it's a bug, because the user isn't
supposed to `repair' intact articles. ;-)

So the question is: Should gnus-summary-repair-multipart modify files
on disk or only modify the buffer?

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo--- PGP key available via WWW   http://rsteib.home.pages.de/