in some cases, plan 9's coincidental inability to run modern programs that do 
unpredictable and undesirable things is a useful feature. mothra, for example, 
doesn't even handle many html tags, but it also doesn't execute unknown 
server-supplied code on my terminal. how can i be sure? because the program is 
small enough to read and understand, and, having done so, i can be reasonably 
certain that it contains no code to do so. quite aside from having the 
functions accidentally or surreptitiously enabled, the functions simply don't 
exist. with most modern "useful" programs (and their dependencies), 
understanding the code isn't a valid approach to security, because your 
lifetime is too short a span to read -- much less comprehend -- the contents of 
the source directory. this is compounded by numerous and constant revisions to 
already unreadably massive piles of code.

what does a given useful program do? who can really say?

harvey seems interesting, but its main objective seems inextricably tied to 
throwing the strength of plan 9's simplicity and relative isolation out the 
window.

sl



On Jul 27, 2015, 10:34 AM, at 10:34 AM, Charles Forsyth 
<[email protected]> wrote:
>On 27 July 2015 at 15:19, Anthony Sorace <[email protected]> wrote:
>
>> (for many, it’s pretty
>> much just a browser)
>>
>
>One of the reasons mere POSIX isn't enough is that there are many
>non-POSIX
>tendrils that have worked their way throughout the system,
>notably d-bus and now systemd, but there are many others, and the "just
>a
>browser" has started to interact with all of them.
>https://code.google.com/p/chromium/issues/detail?id=388628

Reply via email to