Driver to drive?

[Long Hair]

Phil Hobbs wrote:

> On 02/21/2018 04:53 PM, Jim Thompson wrote:

snip

I kissed off buying Ford products in 1977... nevermore.

...Jim Thompson

They're pretty good at the moment--I've bought two in the last 18 months
and couldn't be happier with them.

Cheers

Phil Hobbs

I'll bet Jim bought a truck in which Ford *still* used that same
windshield gasket sealant that runs down your windshield when it gets
hot.

Even after they knew about it, they used it for years. They must have
bought a few too many rail tanker cars full of it to eat it that badly.

I think they actually used it right up through the '80s even.

I am a Chevy / GM man all the way. Always was, always will be.

My first engine rebuild was a BBC 427 Rat motor. That was in 1971.
Just another of the things I can do well.

 

[Long Hair]

Les Cargill wrote:

> bitrex wrote:

snip

Spectre/Meltdown happened because Intel ended up on one end of a
tradeoff. They favored performance over security.

Windows isn't shitty because of the techniques used to develop modern
software, it's mostly because it is and always has been a slave to the
past and backwards compatibility. It's always needed a ground-up rewrite
since about 1990 which it never got.


Windows is like it is because Microsoft only hired fresh-outs and
worked 'em like rented mules.

All things considered - *it's not that bad*. It's slow coding to the
Win32 interface, but it's not horrible.

I remember watching a 133 MHz PC running BeOS display 32 full-motion
videos in windows on its desktop while you could simultaneously edit a
document in the word processor with no hiccups or lag at all.


But you couldn't buy a BeOS box at Best Buy ( or Bob's Computer Store
before that ).

In 1997.
Win 95 would choke and crash under a tenth of that workload. But BeOS
couldn't run 10 year old DOS productivity software. BeOS is gone with
the snows of yesteryear.

Nowadays, one could be ripping a bluray disc into a 35GB mkv file
while at the same time watching a 4GB bluray rip of a movie and
decoding the audio into 7.1 channels, while your active or live desktop
wallpaper does weird shit in the background... without a hiccup.
Next disc... next file... resume movie. Where's my beer...

I had a BeOS box that I used to get into the 99th percentile at seti
at home back in the days before GPUs did it. I had five PCs in my
living room at one time and many of them multi-booted about three to
five different OSes. My dual CPU AMD ran it best from within a Linux
session though. Those were the days... '99... 2000... etc.

 

[Long Hair]

bitrex wrote:

snip

Unfortunately for most of its history x86 wasn't particularly amenable
to fast virtualization, a sleek new OS sold in stores in the late 1990s
which could also run the majority of legacy DOS and Windows software
with little performance hit could have been a contender.

Once OS/2, etc. were gone and Microsoft had the market locked down by
the early 2000s Intel and AMD got crackin' on the hardware
virtualization support like gangbusters, interesting that.

I even had a DesqViewX machine. Shame it is tied to quarterdecks
memory manager or we could still run it. It did real distributed
processing pretty well for what it was. Only if you had TCP/IP protocol
on your net. Or like ours... we had super expensive Thomas Conrad NICs
across 40 Nodes and back then that was not cheap.

I had an OS/2 machine as well, and was even getting ready to become an
OS/2 CNE until MicroSoft took a big dump on IBM.

 

On Wed, 21 Feb 2018 23:52:45 -0500, bitrex
<bitrex@de.lete.earthlink.net> wrote:

On 02/21/2018 10:22 AM, Long Hair wrote:
bitrex wrote:

snip

Someone should design a security system that when e.g. an infrared beam
on the front stoop is broken it paints a half-dozen red dots on the
person standing there from strategically placed lasers all around the
property it's unobtrusive yet probably attention-getting

Until the one that points up at the overhead aircraft gets you some
federal charges.

Remeber those little disc guns we used to shoot at each other?

I would place a simple "slot" on the side of the doorway (or top) with
the gun inside, behond the slot. Very sharp or star tipped discs could
then be launched at any swing angle and folks... sharp discs go right
through Kevlar armor, and many other things.

So, an unsuspecting victim would not even know that he was staring
death right in the face.

There are no door-to-door salesmen in Colorado. Why? Because the
owner can shoot your ass as soon as you cross the perimeter of his
property, much less waiting till you get to the door. Maybe that fear
alone is a good security upgrade.

Still didn't help folks like Versace or John Lennon


Is door-to-door sales still a thing? Can't say I've ever encountered one
knocking on my door in Massachusetts for as long as I've been old enough
to have my own place (15 years or so.)

Except for girl scout cookies/high school fundraisers, maybe

and JW

 

[bitrex]

On 02/22/2018 05:27 PM, Long Hair wrote:

Les Cargill wrote:

bitrex wrote:

snip

Spectre/Meltdown happened because Intel ended up on one end of a
tradeoff. They favored performance over security.

Windows isn't shitty because of the techniques used to develop modern
software, it's mostly because it is and always has been a slave to the
past and backwards compatibility. It's always needed a ground-up rewrite
since about 1990 which it never got.


Windows is like it is because Microsoft only hired fresh-outs and
worked 'em like rented mules.

All things considered - *it's not that bad*. It's slow coding to the
Win32 interface, but it's not horrible.

I remember watching a 133 MHz PC running BeOS display 32 full-motion
videos in windows on its desktop while you could simultaneously edit a
document in the word processor with no hiccups or lag at all.


But you couldn't buy a BeOS box at Best Buy ( or Bob's Computer Store
before that ).

In 1997.
Win 95 would choke and crash under a tenth of that workload. But BeOS
couldn't run 10 year old DOS productivity software. BeOS is gone with
the snows of yesteryear.

Nowadays, one could be ripping a bluray disc into a 35GB mkv file
while at the same time watching a 4GB bluray rip of a movie and
decoding the audio into 7.1 channels, while your active or live desktop
wallpaper does weird shit in the background... without a hiccup.
Next disc... next file... resume movie. Where's my beer...

I had a BeOS box that I used to get into the 99th percentile at seti
at home back in the days before GPUs did it. I had five PCs in my
living room at one time and many of them multi-booted about three to
five different OSes. My dual CPU AMD ran it best from within a Linux
session though. Those were the days... '99... 2000... etc.

In college in the mid-late 1990s as students we had more-or-less two
choices as to what to use, Windows or Mac. It was an art school and
people were trying to do video editing and short film production on
Quadra 900s and Performa pizza-boxes.

Sometimes it worked out OK but that period was kind of painful from a
multimedia/graphics perspective; software was really trying to do too
much with too little compute horsepower to throw at it, and crashes and
data loss were commonplace. Later on they started getting PowerPC based
G3 and G4 Macs which helped a lot.

Sometime around 1998/99 this one guy sprung for a VooDoo2 graphics
accelerator card for his Windows box and everyone shit them selves
oohing and ahhing at the whiz bang effects

File servers, network gateways, and machines for "real shit" were mostly
Solaris/SunOS based with maybe a few Debian Linux machines scattered
around later on. Windows 95/98 machines were for word processing.

As late as 1997 it was still common for incoming students to bring Mac
SEs, Mac IIs, and 386-based Windows 3.1 desktops with them from home for
personal use

 

[John Larkin]

On Thu, 22 Feb 2018 13:44:54 -0800, Robert Baer
<robertbaer@localnet.com> wrote:

..since at least Eagle ver 5 if not earlier.
Did you know that the produced Gerbers in cases DO NOT reflect what
you see on the screen?

Now, someone pointed out that Gerbers are vector.
True,but that has noting to do with fonts or anything else except the
Gerber strokes as a collection, are supposed to represent the original:
a PCB pad, a logo scribble, Cyrilic font, Arabic font, or any other font
AS CREATED AND SEEN ON THE SCREEN.

MOST irritating is that any text in proportional font on the screen
is rendered as their vector font.

Gerbers ARE vectors. Maybe the aperature list could be tweaked to
improve character resolution.


--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Paul]

Robert Baer wrote:

..since at least Eagle ver 5 if not earlier.
Did you know that the produced Gerbers in cases DO NOT reflect what
you see on the screen?

Now, someone pointed out that Gerbers are vector.
True,but that has noting to do with fonts or anything else except the
Gerber strokes as a collection, are supposed to represent the original:
a PCB pad, a logo scribble, Cyrilic font, Arabic font, or any other font
AS CREATED AND SEEN ON THE SCREEN.

MOST irritating is that any text in proportional font on the screen is
rendered as their vector font.

Next, that the (vector font) characters look stunted; most especially
the "R" looks as if its legs were sawed off.
The screen version is elegant; you will NOT get what you see.

Also, some of the vector font symbols are totally trashed, for example
the Registered Trademark symbol ÂŽ looks like a rotated "L" and there is
no replacement unless one wishes to construct one using a circle and (at
least on the screen) a "R"; result has the inelegant and trashy looking
vector font "R".

Such junk is not seen unless one uses an independent Gerber viewer.
**

Now, on occasion, a layer may "inherit" AND/OR "ghost" something from
another layer, and move the position as well.
In one particular case, (so to speak) without asking, a text got
solder-masked (the "ghost" because i never saw it since i do not bother
turning that layer on). And that solder mask was (o be polite)
mis-positioned.
At the same time, a placed circle (layer 1 Top, made for no hole in
center) got duplicated in layer 30 Bstop and position was changed (here
is the "ghosting").
Furthermore, that layer 1 circle got duplicated in layer 290 Tstop and
the position was ALSO changed.
The result was a total mess
**
SO, VERY carefully review ALL layers on the screen at minimum and fix
as needed.
Use a Gerber viewer to double-check the results, especially fonts.

Naturally, because a rich owner of Eagle has decided to also be
greedy, expect these problems WILL NEVER be fixed.

Gerber is back from the days of "Photoplotters", "wheels", "apertures",
"flashes", and photosensitive media (color keys, photoplots).

http://forums.parallax.com/discussion/download/117209/Apeture_1.jpg

Given how crude the tech is, why would you be particularly upset ?

I think it's time for a little SparkFun.

https://learn.sparkfun.com/tutorials/designing-pcbs-advanced-smd

Paul

 

[boB]

On Thu, 22 Feb 2018 13:44:54 -0800, Robert Baer
<robertbaer@localnet.com> wrote:

..since at least Eagle ver 5 if not earlier.
Did you know that the produced Gerbers in cases DO NOT reflect what
you see on the screen?

Now, someone pointed out that Gerbers are vector.
True,but that has noting to do with fonts or anything else except the
Gerber strokes as a collection, are supposed to represent the original:
a PCB pad, a logo scribble, Cyrilic font, Arabic font, or any other font
AS CREATED AND SEEN ON THE SCREEN.

MOST irritating is that any text in proportional font on the screen
is rendered as their vector font.

All of the Gerbers I have created from Eagle (or Pads) appear to be
changed to sweeps/swipes of the gerber aperature. I started at Eagle
4 and went up only to eagle 4.5.

Next, that the (vector font) characters look stunted; most especially
the "R" looks as if its legs were sawed off.
The screen version is elegant; you will NOT get what you see.

Also, some of the vector font symbols are totally trashed, for
example the Registered Trademark symbol Ž looks like a rotated "L" and
there is no replacement unless one wishes to construct one using a
circle and (at least on the screen) a "R"; result has the inelegant and
trashy looking vector font "R".

Such junk is not seen unless one uses an independent Gerber viewer.
**

Now, on occasion, a layer may "inherit" AND/OR "ghost" something from
another layer, and move the position as well.
In one particular case, (so to speak) without asking, a text got
solder-masked (the "ghost" because i never saw it since i do not bother
turning that layer on). And that solder mask was (o be polite)
mis-positioned.
At the same time, a placed circle (layer 1 Top, made for no hole in
center) got duplicated in layer 30 Bstop and position was changed (here
is the "ghosting").
Furthermore, that layer 1 circle got duplicated in layer 290 Tstop
and the position was ALSO changed.
The result was a total mess
**
SO, VERY carefully review ALL layers on the screen at minimum and fix
as needed.
Use a Gerber viewer to double-check the results, especially fonts.

Naturally, because a rich owner of Eagle has decided to also be
greedy, expect these problems WILL NEVER be fixed.

 

[Joerg]

On 2018-02-21 06:50, Long Hair wrote:

Joerg wrote:

On 2018-02-19 15:00, Lasse Langwadt Christensen wrote:
Den mandag den 19. februar 2018 kl. 23.53.49 UTC+1 skrev Joerg:
On 2018-02-19 14:41, Lasse Langwadt Christensen wrote:
Den mandag den 19. februar 2018 kl. 23.13.43 UTC+1 skrev Joerg:
On 2018-02-19 13:34, Clifford Heath wrote:

[...]


I think the LEDs try to detect the dimmer phase angle, and modify
their SMPS setting to draw the desired amount of power. There's
obviously filter time constants in there. The modern dimmers are
doing the same thing, and the control functions fight.


SMPS? In LED light bulbs? That would be like having gold-plated shafts
on the inside of a car transmission.

https://youtu.be/HNaU76L296Q?t=9m56s



That one flickers quite badly. The ones I disected so far either
contained just passives or sometimes a couple of TO92 devices (rest also
all through-hole). Which surprised me because even my bicycle lights
contain buck converters.

on a bicycle light you don't have a lot of voltage to play with and you
need pretty good efficiency when running on batteries


Yes, it is remarkably efficient. It only gets warm when on 8W, at half
power it stays cool even on a summer day when not riding fast. The mains
powered LED lamps are efficient as well. We used to have a total of 150W
in the living room chandelier which is lit at least 4h/day. Now less
than 30W and nothing becomes hot anymore.

The real test would be to place a kill-a-watt tracking switch in the
line *before* the dimmer/lamp circuit, and see what is actually getting
used at each set point.

Ya well, I am not that interested in whether it's 30W, 27W or 33W. Main
thing is that it is a lot less than incandescent. CFL were not an option
in that chandelier because they didn't make them small enough.

Other than that CFL and LED are fairly similar in efficiency but many
LED bulbs are not dimmable.

--
Regards, Joerg

http://www.analogconsultants.com/

 

[Sylvia Else]

There is a light/vent thing in the shower, so I was thinking it would
be cool if the light were on dim all the time.

https://www.dropbox.com/s/po8e9iku1a2zxt0/P2190702.JPG?raw=1

So I need an LED spotlight that runs at low current; most do. Maybe
the one already there would.

I have some dimmable LED lights in my house, but they can't be dimmed so
very much before they start to flicker.

My hallway is pitch black at night in the absence of artificial
lighting, so I installed a couple of discrete LEDs in bezels

<https://www.rapidonline.com/kingbright-rtf5010-led-bezel-clip-prominent-5mm-55-0260>

run off a small transformer, full wave rectified, but no smoothing (to
eliminate capacitors). They run all the time. Works and looks fine.

Sylvia.

 

[Steve Wilson]

John Larkin <jjlarkin@highland_snip_technology.com> wrote:

On Thu, 22 Feb 2018 13:44:54 -0800, Robert Baer
robertbaer@localnet.com> wrote:

..since at least Eagle ver 5 if not earlier.
Did you know that the produced Gerbers in cases DO NOT reflect what
you see on the screen?

Now, someone pointed out that Gerbers are vector.
True,but that has noting to do with fonts or anything else except the
Gerber strokes as a collection, are supposed to represent the original:
a PCB pad, a logo scribble, Cyrilic font, Arabic font, or any other font
AS CREATED AND SEEN ON THE SCREEN.

MOST irritating is that any text in proportional font on the screen
is rendered as their vector font.

Gerbers ARE vectors. Maybe the aperature list could be tweaked to
improve character resolution.

There is an option in Cadsoft V7.7.0 that might help:

Options->User Interface->Misc->Always Vector Font

The help file states:

always displays texts in drawings with the builtin vector font, regardless
of which font is actually set for a particular text

I don't know if this option exists in the Autocad version.

 

On Fri, 23 Feb 2018 11:47:02 +1100, Sylvia Else
<sylvia@not.at.this.address> wrote:

There is a light/vent thing in the shower, so I was thinking it would
be cool if the light were on dim all the time.

https://www.dropbox.com/s/po8e9iku1a2zxt0/P2190702.JPG?raw=1

So I need an LED spotlight that runs at low current; most do. Maybe
the one already there would.

I have some dimmable LED lights in my house, but they can't be dimmed so
very much before they start to flicker.

My hallway is pitch black at night in the absence of artificial
lighting, so I installed a couple of discrete LEDs in bezels

https://www.rapidonline.com/kingbright-rtf5010-led-bezel-clip-prominent-5mm-55-0260

run off a small transformer, full wave rectified, but no smoothing (to
eliminate capacitors). They run all the time. Works and looks fine.

We've had a bunch of LED nightlights running for at least ten years.
They run all the time and work like a champ. I think they were a
couple of bucks each.

 

[George Herold]

On Thursday, February 22, 2018 at 4:45:08 PM UTC-5, Robert Baer wrote:

..since at least Eagle ver 5 if not earlier.
Did you know that the produced Gerbers in cases DO NOT reflect what
you see on the screen?

Now, someone pointed out that Gerbers are vector.
True,but that has noting to do with fonts or anything else except the
Gerber strokes as a collection, are supposed to represent the original:
a PCB pad, a logo scribble, Cyrilic font, Arabic font, or any other font
AS CREATED AND SEEN ON THE SCREEN.

MOST irritating is that any text in proportional font on the screen
is rendered as their vector font.

Next, that the (vector font) characters look stunted; most especially
the "R" looks as if its legs were sawed off.
The screen version is elegant; you will NOT get what you see.

Also, some of the vector font symbols are totally trashed, for
example the Registered Trademark symbol ÂŽ looks like a rotated "L" and
there is no replacement unless one wishes to construct one using a
circle and (at least on the screen) a "R"; result has the inelegant and
trashy looking vector font "R".

Such junk is not seen unless one uses an independent Gerber viewer.
**

Now, on occasion, a layer may "inherit" AND/OR "ghost" something from
another layer, and move the position as well.
In one particular case, (so to speak) without asking, a text got
solder-masked (the "ghost" because i never saw it since i do not bother
turning that layer on). And that solder mask was (o be polite)
mis-positioned.
At the same time, a placed circle (layer 1 Top, made for no hole in
center) got duplicated in layer 30 Bstop and position was changed (here
is the "ghosting").
Furthermore, that layer 1 circle got duplicated in layer 290 Tstop
and the position was ALSO changed.
The result was a total mess
**
SO, VERY carefully review ALL layers on the screen at minimum and fix
as needed.
Use a Gerber viewer to double-check the results, especially fonts.

Naturally, because a rich owner of Eagle has decided to also be
greedy, expect these problems WILL NEVER be fixed.

I try to always check the gerbers with some other viewer
before sending them off.
(I forgot to do that on a recent board :^(
measure twice cut once.

George H.

 

[George Herold]

On Thursday, February 22, 2018 at 6:50:44 PM UTC-5, boB wrote:

On Thu, 22 Feb 2018 13:44:54 -0800, Robert Baer
robertbaer@localnet.com> wrote:

..since at least Eagle ver 5 if not earlier.
Did you know that the produced Gerbers in cases DO NOT reflect what
you see on the screen?

Now, someone pointed out that Gerbers are vector.
True,but that has noting to do with fonts or anything else except the
Gerber strokes as a collection, are supposed to represent the original:
a PCB pad, a logo scribble, Cyrilic font, Arabic font, or any other font
AS CREATED AND SEEN ON THE SCREEN.

MOST irritating is that any text in proportional font on the screen
is rendered as their vector font.



All of the Gerbers I have created from Eagle (or Pads) appear to be
changed to sweeps/swipes of the gerber aperature. I started at Eagle
4 and went up only to eagle 4.5.

I'm not sure what sweeps/swipes are, but I'm still using eagle 4.5.
(KiCad is somewhere in my future. I played with it some, but it's
always faster to use the software you know.)

George H.

Next, that the (vector font) characters look stunted; most especially
the "R" looks as if its legs were sawed off.
The screen version is elegant; you will NOT get what you see.

Also, some of the vector font symbols are totally trashed, for
example the Registered Trademark symbol ÂŽ looks like a rotated "L" and
there is no replacement unless one wishes to construct one using a
circle and (at least on the screen) a "R"; result has the inelegant and
trashy looking vector font "R".

Such junk is not seen unless one uses an independent Gerber viewer.
**

Now, on occasion, a layer may "inherit" AND/OR "ghost" something from
another layer, and move the position as well.
In one particular case, (so to speak) without asking, a text got
solder-masked (the "ghost" because i never saw it since i do not bother
turning that layer on). And that solder mask was (o be polite)
mis-positioned.
At the same time, a placed circle (layer 1 Top, made for no hole in
center) got duplicated in layer 30 Bstop and position was changed (here
is the "ghosting").
Furthermore, that layer 1 circle got duplicated in layer 290 Tstop
and the position was ALSO changed.
The result was a total mess
**
SO, VERY carefully review ALL layers on the screen at minimum and fix
as needed.
Use a Gerber viewer to double-check the results, especially fonts.

Naturally, because a rich owner of Eagle has decided to also be
greedy, expect these problems WILL NEVER be fixed.

 

[Chris]

On 22/02/2018 12:17 AM, bitrex wrote:

On 02/20/2018 08:44 PM, krw@notreal.com wrote:
On Tue, 20 Feb 2018 20:31:56 -0500, bitrex
bitrex@de.lete.earthlink.net> wrote:

On 02/13/2018 10:21 PM, krw@notreal.com wrote:
On Tue, 13 Feb 2018 19:23:23 -0500, bitrex
bitrex@de.lete.earthlink.net> wrote:

On 02/13/2018 07:11 PM, krw@notreal.com wrote:
On Tue, 13 Feb 2018 18:34:34 -0500, bitrex
bitrex@de.lete.earthlink.net> wrote:

On 02/13/2018 06:08 PM, krw@notreal.com wrote:
On Sun, 11 Feb 2018 19:01:06 -0600, Les Cargill
lcargill99@comcast.com> wrote:

John Larkin wrote:
On Sun, 11 Feb 2018 16:45:33 -0500, bitrex
bitrex@de.lete.earthlink.net> wrote:

On 02/11/2018 11:54 AM, John Larkin wrote:

https://www.amazon.com/Brotopia-Breaking-Boys-Silicon-Valley/dp/0735213534/ref=sr_1_1?ie=UTF8&qid=1518366993&sr=8-1&keywords=brotopia




    From the NYT review:

What happened in the 1960s and 1970s was that the industry was
exploding and was starved for talent. There just weren’t enough
people to do the jobs in computing. So they hired these two
psychologists, William Cannon and Dallis Perry, to come up
with a
personality test to screen for good programmers.

Those men decided, in screening about 1,200 men and 200 women,
that good programmers don’t like people — that they have a
complete disinterest in people. These tests were widely
influential and used at various companies for decades.

Most qualities that make a "good programmer" in 2018 have
little to
do with how ingenious or clever you can make the stuff you type
into the box; compilers are very good at this point, CPU
horsepower
is plentiful and premature optimization is the root of all evil.
It's not easy to write high-performance code but it takes
concerted
effort to write truly poor performing code in most languages.

The most valuable skills probably have to do with a) is your
software based on solid design principles b) how well you can
justify that design/explain your reasoning to others and c) how
easily someone can come in fresh and read your documentation and
understand the principles of the design. Which if you have no
interest in people and little experience interacting with them
socially is going to be a tough row to hoe.

There will always be a place for the seriously schizoid
hotshots,
e.g. physics code for games needs to eek out every bit of
performance. A physics engine coder might be insulted if you
sullied his labor with anything as mundane as _gameplay_.

Also the problem with being totally disinterested in people
is that
you usually end up thinking you're a lot better than you
actually
are, without any metric to judge your performance how can you
semi-objectively judge.

That review also names a recently-me-too-fallen individual that
we once did battle with, but I'm not allowed to make any
disparaging remarks about that.

I am continually amazed by the aggressive and toxic Silicon
Valley culture... which has spread into the worldwide
semiconductor industry.

Software used to be fun and novel but then a lot of the industry
got lame and cynical, like hey let's build a grocery store
list app
for a phone with Bluetooth integration, sell that shit for
$2.99 on
the app store and hope to get bought out by the Big G for $50
mil
in 14 months.

I think a lotta women bailed out on that scene because "on
average"
they don't value the bignum bills as highly vs. not getting your
soul crushed to get 'em. Who can blame 'em, I don't. Seems like
most of the "angel" VC funders in the Valley aren't even tech
educated people anymore they're like MBAs and pundits and famous
bloggers and various Wall Street investment bank weenies
after some
fast bucks



I'm reading "Coders at Work", which is a bunch of interviews with
famous programmers. What's impressive is how ad-hoc they are,
with
no obvious systematic approaches, sometimes bottom-up, sometimes
top-down; they just sort of do it.


Most people who have done something noteworthy enough to be
interviewed probably didn't have a lot of examples to work from.

Top-down/bottom up is a legit thing to mix up, tho. Start
with the risk items.

Or the unknowns.

Many of them have no formal, or
informal, training in CS or EE or anything. I doubt that many
know
what a state machine is.


Unsurprising. Software people are very often blind to
determinism in
general.

FSM are not a widely understood technique in software.
God only knows why. Even the people who know about them
often think they are always equivalent to regular expressions.

I naturally code that way but when I've shown my code to "real
programmers", they say "that's neat"!  To me, it's the obvious
way to
get across the street - one step at a time (with some checks
along the
way).

It's not that FSMs aren't a useful design pattern for some problems,
it's just that the software world has moved on to things like OOP,
functional, and generic programming which are more scalable,
extensible,
and abstract

...even where not appropriate.


Nowadays on large projects the most "appropriate" design pattern is
most
often one that generates a codebase 20 or 30 people can work with
simultaneously and not influence each other too much.  You can't have
everyone involved in a project mucking with the state diagrams and
making changes that break other people's shit, so you have to run
modifications thru some master planner guy, which is a bottleneck and
wastes time. It's too "top down." Doesn't scale well.

In a large project like that, programmers shouldn't be making changes
to *ANY* state machines.  That's the what the architect(s) do.  The
programmers implement, they don't design.  Chip designers don't change
the architecture of a processor, either.  They implement the
architect's design.  That's probably why there is so much shit
software out there.  Too many cooks.


That's called "waterfall design" and has been passe in the software
world for I'd guess 30 years or so. It's brittle, inflexible, too
"top-down", there is no one "architect" or small group of architects who
have a complete God's-eye picture of every single "state" or function or
branch of code in a 100 million-line codebase.

So has defect-free software.  There is a reason Win* is such shit.

Processor design is a different animal it takes years and years to bring
a new architecture from concept to finalized design, the software world
is working under much tighter deadlines and customers _expect_ major
changes to be possible for most of the development cycle.

No, it's exactly the same thing.  One group believes in the quality of
their work.

Giving a client this "the design is finalized and being implemented
according to God's plan by the code monkeys and it cannot be touched"
stuff would be a way to find yourself rapidly out of business

So you think the race to the bottom is a good thing.  I could have
guessed that.


There's a lot of lousy code out there, but there's also more great code
out there than there's ever been in history. Well-designed and
implemented software using "modern" practices is not at all hard to
find. Flawed hardware designed top-down is not at all hard to find, e.g.
Spectre, Meltdown.

Windows isn't shitty because of the techniques used to develop modern
software, it's mostly because it is and always has been a slave to the
past and backwards compatibility. It's always needed a ground-up rewrite
since about 1990 which it never got.

I remember watching a 133 MHz PC running BeOS display 32 full-motion
videos in windows on its desktop while you could simultaneously edit a
document in the word processor with no hiccups or lag at all. In 1997.
Win 95 would choke and crash under a tenth of that workload. But BeOS
couldn't run 10 year old DOS productivity software. BeOS is gone with
the snows of yesteryear.

BeOS was reincarnated as Haiku-OS:

https://www.haiku-os.org/

Looks like it's still reasonably active...
--
Cheers,
Chris.

 

[Long Hair]

Joerg wrote:

snip

> Ya well, I am not that interested in whether it's 30W, 27W or 33W.

The suggestion was more toward insuring that the dimmer itself is not
wasting anything.

Main
thing is that it is a lot less than incandescent. CFL were not an option
in that chandelier because they didn't make them small enough.

Other than that CFL and LED are fairly similar in efficiency but many
LED bulbs are not dimmable.

 

[Long Hair]

Chris wrote:

snip

BeOS was reincarnated as Haiku-OS:

https://www.haiku-os.org/

Looks like it's still reasonably active...

Yeah, and except for lack of good hardware hooks (read drivers) or
accelerated graphics on modern cards...


Well... Just a boot screen to look at, I guess.

 

[Reinhardt Behm]

AT Wednesday 21 February 2018 09:31, bitrex wrote:

That's called "waterfall design" and has been passe in the software
world for I'd guess 30 years or so. It's brittle, inflexible, too
"top-down", there is no one "architect" or small group of architects who
have a complete God's-eye picture of every single "state" or function or
branch of code in a 100 million-line codebase.

Today we have "agile" instead. Which means we don't know where our software
is heading but we do sprints to get there.

--
Reinhardt

 

[legg]

On Thu, 22 Feb 2018 07:11:41 -0800, John Larkin
<jjlarkin@highlandtechnology.com> wrote:

On Sun, 18 Feb 2018 13:12:50 -0500, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 02/18/2018 08:14 AM, Chris Jones wrote:
On 17/02/2018 07:18, bloggs.fredbloggs.fred@gmail.com wrote:
On Friday, February 16, 2018 at 12:51:24 PM UTC-5, John Larkin wrote:
On Fri, 16 Feb 2018 08:57:05 -0800 (PST),
bloggs.fredbloggs.fred@gmail.com wrote:

On Thursday, February 15, 2018 at 8:32:35 PM UTC-5, John Larkin wrote:
LM317's like some ESR in their output capacitors. I don't want any
electrolytic or tantalum caps in my new thing, just ceramics, and the
sim sure rings:

https://www.dropbox.com/s/11b3w42nsvpliki/317_nocomp.jpg?raw=1

But this fixes it:

https://www.dropbox.com/s/9q80heyfbwh5frp/317_comp.jpg?raw=1

This ain't rocket science, but I haven't seen it done before.

317 needs no such ESR compensation.

The data sheet says it does.

The ringing looks suspiciously like excitation of the SRF of an
output capacitor.

The frequency is low, and is different on the rising and falling edges
of the load current pulse. It's the chip pseudo-inductance resonating,
not the cap's ESL. If the ringing were local to the caps, my damping
on ADJ wouldn't fix that.

Cap series L makes a different waveform than paralleled inductance.


Did your model give it any ESL? And your solution merely reduces the
shunt resistance by a factor of 20x which probably has more to do
with damping than anything else.

With a big cap from ADJ to ground, it rings badly, too. It has to be
the *right* capacitor to damp the ringing.

I tried this with two different LM317 models; the ringing is somewhat
different (the LT317 is better), but the damping idea is the same.

It's amazing that LT ever made a 317. I think they did that early on,
when they needed some revenue. They want $4 for it! I'm paying less
than a tenth of that for TI.

I doubt you're going to see this energetic resonance on anything other
than the LT part.


I doubted it too, but found out the hard way when:
my 337's all oscillated, and
the 317s rang so badly that the oscillation ripple on the positive rails
was even bigger than on the negative rails.

The 317s wouldn't oscillate by themselves, but they would ring like a
bell even after I cured the 337's of oscillation.

I had to scratch off a lot of solder mask and tack on many tantalums to
cure my boards. Quite embarassing.


Check out the Erroll Dietz article I posted upthread.

He used three values of Cadj, 0, 10u, and 1000u. He didn't try
anything like 22nF. I'm sort of surprised that nobody seems to have
tried that, or at least publicized it.

As I recall, from the time when I was actually checking transient
response and output noise of commercial linear prototypes, the kudos
for getting a 10n cap to do the job that a 10uF part was illustrated
to do (or not to do) in the literature, wasn't worth mentioning.

If it was, it was as the prelude to the inevitable 'Why not just leave
it out? Nobody's going to do that anyways.'

Tantalum caps were not even a consideration.....but the product still
worked over the temperature range.

With the harnessing involved, critical decoupling was always at the
load, so my measurements on the output terminals were just 'nice to
know' ~ required for a test spec or sales blurb.

Power supplys were, and are still, just not sexy.

RL

 

[boB]

On Thu, 22 Feb 2018 17:56:32 -0800 (PST), George Herold
<gherold@teachspin.com> wrote:

On Thursday, February 22, 2018 at 6:50:44 PM UTC-5, boB wrote:
On Thu, 22 Feb 2018 13:44:54 -0800, Robert Baer
robertbaer@localnet.com> wrote:

..since at least Eagle ver 5 if not earlier.
Did you know that the produced Gerbers in cases DO NOT reflect what
you see on the screen?

Now, someone pointed out that Gerbers are vector.
True,but that has noting to do with fonts or anything else except the
Gerber strokes as a collection, are supposed to represent the original:
a PCB pad, a logo scribble, Cyrilic font, Arabic font, or any other font
AS CREATED AND SEEN ON THE SCREEN.

MOST irritating is that any text in proportional font on the screen
is rendered as their vector font.



All of the Gerbers I have created from Eagle (or Pads) appear to be
changed to sweeps/swipes of the gerber aperature. I started at Eagle
4 and went up only to eagle 4.5.

I'm not sure what sweeps/swipes are, but I'm still using eagle 4.5.
(KiCad is somewhere in my future. I played with it some, but it's
always faster to use the software you know.)

George H.

I meant the X,Y start X,Y end points also sweep out the text fonts
instead of actually placing the ascii text at a certain spot in the
gerber and interpreted by the gerber viewer as text.

Have you looked inside the gerber files and seen a font name mentioned
near the top of the file ? I could imagine problems arising from
that method if it is possible. Never seen it myself though. Just
lines and dots of various widths making up everything in the gerber
output.

boB

Next, that the (vector font) characters look stunted; most especially
the "R" looks as if its legs were sawed off.
The screen version is elegant; you will NOT get what you see.

Also, some of the vector font symbols are totally trashed, for
example the Registered Trademark symbol Ž looks like a rotated "L" and
there is no replacement unless one wishes to construct one using a
circle and (at least on the screen) a "R"; result has the inelegant and
trashy looking vector font "R".

Such junk is not seen unless one uses an independent Gerber viewer.
**

Now, on occasion, a layer may "inherit" AND/OR "ghost" something from
another layer, and move the position as well.
In one particular case, (so to speak) without asking, a text got
solder-masked (the "ghost" because i never saw it since i do not bother
turning that layer on). And that solder mask was (o be polite)
mis-positioned.
At the same time, a placed circle (layer 1 Top, made for no hole in
center) got duplicated in layer 30 Bstop and position was changed (here
is the "ghosting").
Furthermore, that layer 1 circle got duplicated in layer 290 Tstop
and the position was ALSO changed.
The result was a total mess
**
SO, VERY carefully review ALL layers on the screen at minimum and fix
as needed.
Use a Gerber viewer to double-check the results, especially fonts.

Naturally, because a rich owner of Eagle has decided to also be
greedy, expect these problems WILL NEVER be fixed.