Derek Martin <inva...@pizzashack.org> wrote:
> On Sat, Jan 09, 2021 at 10:54:28PM +0000, Eric Wong wrote:
> > Hi Remco,
> > 
> > So I'm looking at 9da4e6e11e7037668d0ca7e8f5d6773d26e379ac
> > (I noticed this in mutt 2.0.2 on FreeBSD)
> > 
> > This is a bad change
> 
> You should have just stopped there.  This is a bad change, as I argued
> when it was originally discussed.  Given that there was neither
> agreement about how to change it, even amongst those who thought it
> should change, nor any concrete example of harm caused by the original
> format, it should never have been changed.  

I admit I don't pay very close attention to this list since mutt
mostly works well for me; but vaguely remembering it was rejected
last year...

(Maybe it was discussed on GitLab, but I can't be bothered with
 swap storms on JS/CAPTCHA-infested sites like GL)

> This thread is a good example of why you don't arbtrarily change
> things that have worked the same way for 20 years without a very good
> reason:  You have NO IDEA what the change will break or how many
> people will be affected by the breakage.  For all of the above
> reasons, the new patch should not be accepted, and the original change
> should be REVERTED.

Well, mistakes happen...  How we deal with them moving forward
is important, though.

Fwiw, I concur the PID in Message-IDs was a bad idea since it
does unnecessarily leak information about system usage.  I'm no
security expert, though I seem to remember kernel patches that
would randomize PID allocation in Linux.

> FWIW, the message ID should NOT be URL-encoded.  That is the job of
> the web interface for the archives/whatever.  Using such an encoding
> makes it harder for the humans to read and compare message IDs, but
> programs have no trouble doing such conversions as needed.  The only
> time it should be URL-encoded is when it is in a URL. Mutt doesn't
> generate those.

Right.  In case there's any confusion, my proposed patch does
not do URL-encoding; it merely avoids characters that would need
URL-encoding when a Message-ID gets used as part of a URL.

Reply via email to