UNIX

[Matthew Dillon]

In article <1991May26.212136.20643 at digibd.com> rhealey at digibd.com (Rob Healey) writes:
>In article <7781 at ecs.soton.ac.uk> etj90 at ecs.soton.ac.uk (Khaos) writes:
>>In article <dillon.7891 at overload.Berkeley.CA.US>, dillon at overload.Berkeley.CA.US (Matthew Dillon) writes:
>>>	I believe the best bet will be BSD.  Current rumor puts it at about a
>>>	year away. You can *bet* that the moment they release it I will begin
>>>	to port it to the Amiga.  BSD has always been a much cleaner
>>>					    -Matt
>
>	Ummm, I don't know about BSD being CLEANER, just first... I
>	think non-english speaking UNIX people might have a problem
>	with BSD as well. And it would take a REAL good argument to
>	convince me that BSD sockets and Co. is cleaner than STREAMS
>	and TLI or as flexible when it comes to code using multiple
>	network types dynamically without recompilation.
>
>	I grant youthat MOST network code is written to the BSD socket
>	model and has a bent toward BSD OS model because BSD had
>	good OS extentions before SV. I think it's QUITE a bit of a stretch
>	to call BSD cleaner than R4 tho.
>
>	Flames and religeous wars to e-mail, c.u.a doesn't need
>	religous wars filling it's bandwidth...
>
>		-Rob

    Hm.. well, note that I never said BSD was better, just cleaner.  From
    what I hear on this group, AT&T *really* rushed R4.  I am *not* a Sys
    VR4 user myself, my impressions are gained from conversations. However,
    I did hack at the BSD kernel while at Berkeley (as well as write tons
    of user level stuff).

    The only two major items that BSD lacks are intelligent device control
    (i.e. filesystem-mapped in a meaningful way) and shared memory.  Other
    items such as semaphores & signalling are also lacking but not quite as
    important.

    The socket stuff is not so difficult as all that, just a bit
    overblown... they created interface calls with an order of magnitude
    too much flexibility which tends to require a lot of work to get right.

    One assumes that 4.4 will address these issues.

    As far as language problems go, while I agree that internationalization
    is becomming a greater necessity these days, I do not believe that it
    is an kernel core issue, but more a systems level issue (e.g. shared
    library of terms, ...)

					-Matt

>--
>
>Rob Healey					     rhealey at digibd.com
>Digi International (DigiBoard)
>Eden Prairie, MN				     (612) 943-9020

--

    Matthew Dillon	    dillon at Overload.Berkeley.CA.US
    891 Regal Rd.	    uunet.uu.net!overload!dillon
    Berkeley, Ca. 94708
    USA