[RndTbl] Linux "write"

Gilbert E. Detillieux gedetil at cs.umanitoba.ca
Wed Jul 5 14:45:14 CDT 2017

On 03/07/2017 2:29 AM, Trevor Cordes wrote:
> On 2017-07-03 Adam Thompson wrote:
>> Sorry, I have to agree with them.  Your use case is... pretty much
>> obsolete, at least in my opinion.
> Getting philosophical: one of the virtues of *NIX I always extol to
> outsiders and insiders alike is how *NIX always builds on the past, and
> rarely (until recently) destroys it.  This was its strength.  It didn't
> break %#!@.  It didn't revamp itself into something new we hate every 5
> years (ehem, Micro$oft).  It just kept adding features.
> I would say that was true from when I started on *NIX in 1992 through
> to about 2010.  During that heyday, almost nothing I cared about or used
> ever broke.

Ah, but that was just after the Great UNIX Standards Wars of the 
'80's...  In the early years of UNIX, things were a bit more fluid and 
new versions did introduce incompatibilities.  (And the original 
creators of UNIX moved on to other, rather different and incompatible OS 
projects, like Plan 9.)

What the competing standards did accomplish is (at least for a while) an 
awareness of the importance of software portability and trying to nail 
down a minimal set of compatibility standards (like POSIX, at least, if 
you didn't want to commit to either a SYSV- or BSD-compatible system).

Then Linux came along, and (at least in the early years) took the 
attitude that features from any of the various UNIX flavours that were 
useful were all up for grabs, and they took the best and left the rest. 
(Or, arguably, the other way around, especially if you're a hard-core 
BSD fan.)  As long as a lot of the Linux developers had one foot in 
another camp, it kept things in the API's _reasonably_ compatible with 
other UNIX flavours.

But that was in the past...

> Since then it's changed drastically. I used to look
> forward to the new RH (later Fedora) version release, now I dread every
> latest Fedora, because I know it will break 2-4 things that will cost
> me 2-4 days (or more, sometimes weeks) debugging or working around.
> Somewhere along the way the dev mindset changed from respecting the past
> to actively destroying it for destruction's sake.  In 1992 I respected
> what the guys in 1970 did, so why can't the 2017 devs respect what we
> did in 1995?
> If you change this aspect of *NIX, then you knock down one of its key
> pillars.  It becomes another Windows where you throw away your mastery
> and tools every 5-10 years; never really getting ahead, just keeping up.

I think a lot of the Linux development lost its way years ago, when they 
focused far too much on the Desktop (in a vain bid for "world 
domination") and ignored the more established server market which had 
been their bread and butter.  The longer they do that, the more they'll 
continue to alienate the server market, who'll move to other platforms 
(like *BSD).  And in the end, they're not going to win the desktop 
market, but will always be trying to catch up to the big boys.

>> real world example of https://xkcd.com/1172/ from my perspective.
> Ha, well, not quite... wall and write and utmp are documented things
> that are supposed to work as intended (and have for, 40 years?).  But
> good cartoon anyhow :-)

As far as write(1) and other chat systems go, this is perhaps more to 
the point...


As a very early user of write(1), I had already grown to hate it long 
ago, for social/usage reasons more than technical issues, so I don't 
particularly care if it goes.  And it can take the rest of the 
proprietary chat systems along with it!  (Never had much use for any of 
them, and hate the fact that they're all busy reinventing the same 
broken wheel in their own incompatible way, rather than working toward a 
unifying standard that is actually useful.)

Call me old-fashioned, but gimme an IMAP/SMTP-based e-mail client over 
the above, any day!  Even the very limited (and limiting) iOS Mail app 
is more versatile and useful than any chat app I've ever had to struggle 

>> Anyway, I don't have a neat solution wrapped up for you, sorry.  I
>> finally gave up on Linux as a desktop last year, and I was using
>> gnome terminal to the bitter end (and not caring whether it wrote to
> Shame, as I think Linux as a desktop is finally "here".  When I can
> convert my wife, mom and dad to Linux without any pushback or non-stop
> support calls, you know Linux has arrived.  In fact, I deal with far
> less "support" with them on Linux than I ever did with Windows!  And I
> can fix it all with ssh instead of vnc.  (A funny aside: I tried some
> on stock Fedora GNOME 3/shell and they couldn't figure anything out;
> but switching to XFCE solved that problem.  Heck, I can barely figure
> out GNOME 3, so I'm not sure who their target market is.)

See my desktop/world-domination rant above.  :)

Besides, Gnome 3 is to GUI's as "newspeak" is to language.  (Read "1984" 
if you haven't already.)

> Personally, I can't see how anyone could use Windows for serious
> multitasking desktop work these days... but that's a whole other
> discussion.

As far as using Windows, I'm going to stick with Win7 for as long as I 
can.  Really don't like what they've done with the GUI in a lame attempt 
at being more tablet-friendly.  Not particularly fond of the macOS 
development direction either, but it's not quite as ugly and broken as 
Windows 10.

Gilbert E. Detillieux		E-mail:	<gedetil at cs.umanitoba.ca>
Dept. of Computer Science	Web:	http://www.cs.umanitoba.ca/~gedetil/
University of Manitoba		Phone:	(204)474-8161
Winnipeg MB CANADA  R3T 2N2	Fax:	(204)474-7609

More information about the Roundtable mailing list