On 12/22/2020 0:44, Hans-Bernhard Bröker wrote:
> Am 21.12.2020 um 16:29 schrieb Dimiter_Popoff:
> 
>> Not so. Clearly you have never designed in ATA drives.
> 
> The level of your arrogance never disappoints.

Sorry, did not mean to be arrogant. I can see it can be interpreted
like this though, I must have been thinking on all the history I have
had with ATA drives, CPLD interfaces and drivers I have been doing for
them. Not very interesting or complex to do but quite bulky.
Sorry again if I offended you.

Dimiter

Am 21.12.2020 um 16:29 schrieb Dimiter_Popoff:

> Not so. Clearly you have never designed in ATA drives.

The level of your arrogance never disappoints.

On 12/21/2020 0:56, Hans-Bernhard Bröker wrote:
> Am 20.12.2020 um 22:23 schrieb Dimiter_Popoff:
> 
>> But for people familiar with ATA and subsequently SATA and in the
>> context above "master" and "slave" refer to device 0 and device 1
>> on the ATA cable, 
> 
> Not really.  Parallel ATA learnd this ("cable select") as an option long 
> after the master/slave roles had been assigned.
> 
> The primary choice was made by jumpers on the devices, with three 
> options: master, slave, or cable-select.  Only if both devices were 
> jumpered CS would their position on the cable have any significance.
> 
> The master/slave terminology was attached because device 1 was unable to 
> work on its own.  I.e. a single device would always have to be device 0.
> 

Not so. Clearly you have never designed in ATA drives.

The "cable select" signal is just a pin connected to the host and to
one of the two devices, thus enabling them to know which of the two is
device 0 and which is device 1 (lookup d1153r18, it is explained well
enough there).
The jumpers on the devices are to override the cable select connection
and are unnecessary on correctly designed hosts/devices.

And of course you can have a single device on an ATA cable which
is device 1, although it is rarely done and typically unnecessary.

Dimiter


======================================================
Dimiter Popoff, TGI             http://www.tgi-sci.com
======================================================
http://www.flickr.com/photos/didi_tgi/

Am 20.12.2020 um 22:23 schrieb Dimiter_Popoff:

> But for people familiar with ATA and subsequently SATA and in the
> context above "master" and "slave" refer to device 0 and device 1
> on the ATA cable, 

Not really.  Parallel ATA learnd this ("cable select") as an option long 
after the master/slave roles had been assigned.

The primary choice was made by jumpers on the devices, with three 
options: master, slave, or cable-select.  Only if both devices were 
jumpered CS would their position on the cable have any significance.

The master/slave terminology was attached because device 1 was unable to 
work on its own.  I.e. a single device would always have to be device 0.

On 12/20/2020 20:17, Grant Edwards wrote:
> On 2020-12-20, Dimiter_Popoff <dp@tgi-sci.com> wrote:
> 
>> SATA does just a single device per link so no "master/slave" to
>> be talked about
> 
> Odd. All my SATA cables have _two_ ends, and there's a device at each
> end. I was under the distinct impression that at one end is the master
> and at the other end there's a slave.
> 
> There are are tons of other point-point protocols that use the
> master/slave nomenclature, even if SATA doesn't.
> 
> --
> Grant
> 

Hah, you may call them that of course, some people may even be
calling them that.

But for people familiar with ATA and subsequently SATA and in the
context above "master" and "slave" refer to device 0 and device 1
on the ATA cable, you probably remember your parallel ATA (aka IDE)
cables having 3 connectors: one for the host, one for device 0
(which some called "master") and one for device 1 (which some
called "slave"). Actually I don't know where that master/slave
came from, perhaps from the times ATA was called IDE which was before
I used any ATA drives in our products (used SCSI back then, until
they stopped at 810MB or so for 2.5" drives).

Dimiter

======================================================
Dimiter Popoff, TGI             http://www.tgi-sci.com
======================================================
http://www.flickr.com/photos/didi_tgi/

On 2020-12-20, Dimiter_Popoff <dp@tgi-sci.com> wrote:

> SATA does just a single device per link so no "master/slave" to
> be talked about

Odd. All my SATA cables have _two_ ends, and there's a device at each
end. I was under the distinct impression that at one end is the master
and at the other end there's a slave.

There are are tons of other point-point protocols that use the
master/slave nomenclature, even if SATA doesn't.

--
Grant

On 12/19/2020 23:41, Kent Dickey wrote:
 > ...
>...  And SATA got rid of master/slave, too.  And so on...
> 

SATA does just a single device per link so no "master/slave" to
be talked about (and the ATA standards talk of device 0 and 1,
no master/slave there either IIRC since > 20 years).

However you can connect two ATA/SATA bridges to an ATA port
which does device 0 and 1.... and device 1 (the "slave") will not
always work just fine, depends on which version of the bridge you
combine for device 1 (slave) with which for device 0 (master)....
That for the only truly working bridge on the market from Marvell.

:D :D

So much talk about that politically (in?) correct nonsense.
Context is enough offend or not offend anyone, the choice of words
is actually irrelevant. Those who want to offend will find ways to do it
using any sort of words and those who want to play offended victims
won't run out of "offensive" words either. Words exist and do have 
meaning, banning a certain word is outright stupid. It is the messages
people convey which can be offensive, not the words.

Dimiter

======================================================
Dimiter Popoff, TGI             http://www.tgi-sci.com
======================================================
http://www.flickr.com/photos/didi_tgi/

Am 19.12.2020 um 03:03 schrieb Brett:

> Master - Apprentice

Doesn't work.  "Apprentice" is not just any member of the master's work 
force; it's a career step towards becoming a master in one's own right. 
That doesn't fit a SPI or I2C device at all.  The correct term from that 
field would be "hand."

And that's before we consider that due to its relation to the past of 
the outgoing POTUS, the word "Apprentice" is just too close to being 
non-PC to bet a concerted world-wide data-sheet audit-and-rewrite on.

The core problem, though, is that however large the bubble of available 
synonyms is, excessive application of PC rules will eventually drain it 
completely.  Eventually you'll be unable to talk about anything, because 
any word actually meaning anything close to what you're trying to say 
will have been found offensive by someone, somewhere, at some point in 
time.  The idea of PC has a risk of being abused as a DDOS attack 
against the system of "language".

> Works where the master can be replaced on failure, which is true with many
> modern distributed software systems.

In which case it was never truly a master-slave system in the first place.

In article <rra4up$okg$1@dont-email.me>,
David Brown  <david.brown@hesbynett.no> wrote:
>On 15/12/2020 10:32, Brad Eckert wrote:
>> It seems the terms *master* and *slave* are being phased out of the
>> tech lexicon with help from Google. I know this is like herding cats,
>> but I have a proposal for a terminology change.
>> 
>> Just replace *master* and *slave* with *Alice* and *Bob*
>> respectively. *Bob* attaches to *Alice* and then *Alice* tells *Bob*
>> what to do, so it's easy to remember. Isn't that more fun than
>> *master* and *slave*?
>> 
>> In SPI terminology, *mosi* and *miso* become *aobi* and *aibo*. If
>> you have multiple Bobs on the bus, it gets better. Alice has a
>> meeting with the Bobs.
>> 
>
>The software world is full of terms that can offend people if people
>choose to be offended.  Just look at processes on *nix systems - you get
>daemons and zombies, you stop a process by killing it, parent processes
>can't die until all their children are dead (or else the children turn
>into zombies), and so on.  I remember someone on my CSP course at
>university complaining about the terms "angelic choice" and "demonic
>choice".

IDE drives are configured as masters and slaves.  All end users had to
deal with setting up drives as "masters" or "slaves".  It's a problem for
tech support.  I don't doubt racist people did racist things with these
terms.

It's not so much that it only affects the small number of people designing
and building machines, it's that the terminology often gets out into the
rest of the world where it can be unnecessarily problematic.  It's a
sign of maturity that the industry sees no reason to go out of its way
to be offensive and childish to end users.

In my software, I have a Monarch which tells Serfs what to do.  It's very
useful since no one else uses those terms, so I always know what it's
referring to.

I wouldn't spend time redefining old standards, but for new standards, just
pick different words.  PCI master/slave became PCIe
initiator/requester/transmitter and completer/receiver.  What's the problem
with doing this?  PCIe is nearly 20 years old now--did anyone even realize
this was being done?  PCIe still refers to the PCI concepts using Master
and Slave--they didn't redefine anything, they just stopped using it for
anything PCIe-only.  And SATA got rid of master/slave, too.  And so on...

Basically, industry has already been doing this for more than two decades.

And the Alice and Bob stuff is like throwing a tantrum.

Kent

Brad Eckert <hwfwguy@gmail.com> wrote:
> It seems the terms *master* and *slave* are being phased out of the tech
> lexicon with help from Google. I know this is like herding cats, but I
> have a proposal for a terminology change.
> 
> Just replace *master* and *slave* with *Alice* and *Bob* respectively.
> *Bob* attaches to *Alice* and then *Alice* tells *Bob* what to do, so
> it's easy to remember. Isn't that more fun than *master* and *slave*?
> 
> In SPI terminology, *mosi* and *miso* become *aobi* and *aibo*. If you
> have multiple Bobs on the bus, it gets better. Alice has a meeting with the Bobs.
> 

Master - Apprentice
Works where the master can be replaced on failure, which is true with many
modern distributed software systems.

Just google synonyms for these words recursively.