Re: [evolution-patches] newest patch for fix non rfc2047 compliant i18n mailer

[Not Zed <notzed ximian com>]

are you sure this is right?  why are the q and b cases different?  shouldn't they be the same?  they should really be the same code instance too, they both do basically the same thing.

and you do a lot of strstr - its fairly expensive to keep doing that over and over the same string.

this also will look for broken strings across whitespace, which is an even more broken case, but this will break plain strings which don't otherwise need encoding.

e.g. what will this do?

   =? foo bar

It will go into the else case, then wont match any of the subcases, it will then start scanning backwards until it finds =? -> which might clearly go beyond the start of the string!  crash!

You should definitly not be scanning backwards at all, if you can't find ?= anywhere going forward then you certainly dont care if you find it goind backwards.

You should also skip things you've already scanned.

And

 =?  foo bar  baz .... ?b? 

will also do weird shit, and

=? foo bar baz ... ?b?  balh blah ?=

Will also break.  I dont think this should be treated as an encoded word, even if some mailers write similar crap out to that (i.e. embedded whitespace).

So still some issues, its more on the right track though.


On Wed, 2004-06-02 at 18:47 +0800, cantona wrote:
> FIXED another case like
> "=?Big5?B?t1Gw3aazw+Swpm1vdXNlpm6k4rdQPyAop9qwpqTitFikaqRVKSAu?==?Big5?B?Li4=?="
>
> I promise I have tested all cases that this patch is working fine
>
> On Wed, 2004-06-02 at 18:00 +0800, cantona wrote:
> > Hi, 
> > 
> > This is the newest patch. Working with no overhead and no broken the
> > valid header. Check this attached patch and please ignore my old post of
> > the patch.
> > 
> > Thanks,
> > Cantona

--

Michael Zucchi <notzed ximian com> Ximian Evolution and Free Software Developer