Mailing-List: contact users-help@spamassassin.apache.org; run by ezmlm
Precedence: bulk
From: "Bill Cole" <sausers-20150205@billmail.scconsult.com>
To: "Kevin A. McGrail" <KMcGrail@PCCC.com>
Cc: SpamAssassin <users@spamassassin.apache.org>
Subject: Re: Cron <smmsp@thebighonker> /home/ler/bin/update-bayes.sh
Date: Sat, 06 Jun 2015 13:55:25 -0400
Message-ID: <15072EF0-90A3-4F98-8635-A5F995218795@billmail.scconsult.com>
In-Reply-To: <79B6AFCA-1E50-46CF-BC49-9FF3C12473EC@PCCC.com>
References: <E1Z0xY8-0009J1-Ux@thebighonker.lerctr.org>
 <8711b404114a3385aaa31db02f5676dd@thebighonker.lerctr.org>
 <5572084B.5000402@PCCC.com>
 <F25BEF2F-B3B9-4D86-89D8-CBC33FDDC251@billmail.scconsult.com>
 <55721CB2.4020609@PCCC.com>
 <1DD79A28-FC7E-4053-9298-3B9FA1136BCE@billmail.scconsult.com>
 <79B6AFCA-1E50-46CF-BC49-9FF3C12473EC@PCCC.com>
MIME-Version: 1.0
Content-Type: text/plain; format=flowed

On 6 Jun 2015, at 6:16, Kevin A. McGrail wrote:

> Well, referencing an rfc is good but please remember rfc 1121 where we 
> can extrapolate that robustness and contravariance overrides because 
> your statements are contrary to the behavior of mua's which DO 
> highlight the urls.
>
> See my screenshot attqched of k9 where all 7 of your bad examples 
> highlighted as links.  I expect similar behavior with other mua's as 
> well.

Sure. As I recall, MS was so impressed with the ancient "email virus" 
hoax that when they wrote Outlook they made sure to make it possible. 
And when Apple saw how horribly Outlook handled format=flowed mail they 
stopped generating it and slavishly replicated Outlook's shoddiness even 
as MS was fixing it.

MUAs are shoddy garbageware, generally worth less than what users pay 
for them. That has it true now and always has been, to at least three 
9's. K-9 is no exception, despite nothing on Android being better when 
last I looked. I understand the rationale that it makes sense to filter 
as a domain anything some crap MUA activates as a link but that approach 
creates unbounded scope creep because there's no limit to how horribly 
risky MUAs can be.

> But hey, while I might disagree with you, the project is open source.  
> Sign a CLA,

Did that bit the last time you suggested it, May 6 (bug 7005). But...

> work up a patch, Add an option to flag only uris prefixed with a 
> protocol, document it well, implement the function where it is need, 
> test it and submit it for inclusion.

My stream of time slices fit for coding doesn't even qualify as a 
stream. I'll be lucky if I can find the right point of attack for this 
this month, no idea when I'd actually be able to write anything worth 
submitting.