The compatability list is stuf they
know works because they tested it.
Other hardware may, but it's like
older Linuxes:
- brand-new stuff won't work yet
- really old stuff probably won't
- and seriously wierd stuff definitely won't (;-))
The Conductor writes: But consumer electronics goes mostly with the dealership model: you take a fritzed Tivo to the place you bought it
True for high-end purchases, but for anything
like a TV or a stereo component, the box contains
a sheet thast says "Do NOT return this to the store
if there is a problem, call 1-800-the-repair-depot".
The phone number leads you to the manufacturer's
helpdesk, who direct you to someone who's a
high-grade certified version of the backyard mechanic.
This doesn't apply to cars (yet!),
but has definitely applied to scooters
and low-end mororcycles, where there
were unauthorized-by-the-vendor sub-dealers in the
products, who send you to the
dealer's repair shop if there's a problem.
Some companies support movement between the "dual ladders" and have positions for very senior engineers. Certainly Siemens had that: my cheif software
architect (hi Russell!) was such, and my current
employer does too.
My former Director at Geac, Jacob Slonim, had a standard policy to keep people engaged, learning and growing in value to both themselves and the company: If you
went for a promotion on the tech ladder, he'd
second you to the business/management side for
at least a quarter. A programmer got to learn what
a business analyst does, an architect gets to
learn team management, and so on.
Net result? Senior engineers with
insight into the business, and sensitivity
about not being "the unmanagable engineer" (;-))
Other chip design companies have found that some of the
"de rigeur" optimizations either don't buy
anything or cost so much silicon that
applying them forces the designer to pessimize something else.
An example I know about from an old version of Samba is that
using branch prediction to compile a binary
that expects all debugging code to be branched-around shows no detectable speed improvement.
It turned out that a complex debug macro expanded to a chunk of code which was bigger than a
cache line on the machine I had. Net result?
The branch, predicted or not, causes a cache refill, whose time completely dominates the time saved/spent by the branch prediction.
To use this optimixation, therefor, you would
need to dedicate more silicon to cache line length in the i-cache and optimize the debug code for size, not speed. An interesting problem in hardware-software co-design (;-))
When I was little, computer programming was
considered a "female" job, akin
to teaching small children to tie their
shoes... Odd that I've not heard
that lately.
digitalgimpus writes: storage is pretty fast already. The average consumer doesn't need any faster.
Average developers tend to notice when their compile
speed drops by a factor of four, because they or
their sysadmin replaced a 4-disk stripe with
a single larger disk.
Even average users might notice if
their programs and data files load four times slower...
It happened to be Canada, but the nature of the
charge is really a side point.
The vendor's lawyer probably could have won
that battle, but be dragged right back into court
by the crown adding a copyright charge, and
would then have lost. So it was better to cop
a plea **before** they pissed off the crown
attorney (the District Attorney in the US).
The important part is that the defence using
the contract failed miserably,and should arguably
fail the same way in the US.
In addition, one might usefully ask if
the EULA, contract or other licence requires
one to break the law, including inadvertantly
joining a criminal conspiracy to keep bugs a secret.
It was prosecuted as theft and posession
of stolen property. Copyright didn't enter
the discussion. From memory, the
vendor plead guilty to posession of stolen
property and was fined.
sjames writes: Actually, that's a strongly contested issue in law right now. Some maintain that because the license is hidden away inside the box, it carrys no force and first sale applies...
Some years ago, a CP/M modem program had a
license which prohibited reverse engineering or
disclosing anything about the program. Inspection
proved that it was in fact a stolen copy of another
modem program.
A purchaser reported this to the police,
who promptly contacted the company that it
had been stolen from, and prosecuted the
thieves. They claimed the reverse engineering
was illegal, but the court ruled that the
contract was void, as it required the ultimate
customer to break the law by keeping the
theft a secret. It was therefor null and
void.
If the anti-virus company was requiring
the users to hide the fact the program was
"not suitable for the purpose sold", then
they arguably were demanding the users
to enter into a criminal conspiracy
in restraint of trade.
Users of such programs may wish to speak to their solicitors, lest by obeying the contract terms they inadvertently commit an offense (;-))
I vividly remember Apple ][ and CP/M disk
protection schemes built on breaking the disk
format, "dongles" plugged into CP/M and DOS
printer ports and all sorts of unsucessful
software schemes.
Each time they made money for the
sellers of the scheme, but harmed the
purchasers. And I don't mean the
end-users, I mean the companies that
shipped software that depended on
unreliable and sometimes deliberately
broken hardware.
Customers couldn't use the products,
and returned them for a refund. Which
made the dealers relctant to stock them,
and eventually the products were supplanted
by their more functional competitors.
My old director (hi, Dean!) made everybody in
a 70-person department take the same one-week project management course, so we could
- communicate with the project managers
- communicate with each other in the same language
- recognize problems the PMs faced, and
- contribute to solving them.
As you might expect, we saw every one of the cited
problems, including pusilanimous
management in the next two years of
work. We did better that most at
surviving, them, though!
Slashdot Mirror
It would serve them right, though! MS loves to fork things to make them Windows-only, like Java -> C#, so turnabout's fair play.
--dave
- brand-new stuff won't work yet
- really old stuff probably won't
- and seriously wierd stuff definitely won't (;-))
--dave
True for high-end purchases, but for anything like a TV or a stereo component, the box contains a sheet thast says "Do NOT return this to the store if there is a problem, call 1-800-the-repair-depot". The phone number leads you to the manufacturer's helpdesk, who direct you to someone who's a high-grade certified version of the backyard mechanic.
This doesn't apply to cars (yet!), but has definitely applied to scooters and low-end mororcycles, where there were unauthorized-by-the-vendor sub-dealers in the products, who send you to the dealer's repair shop if there's a problem.
--dave
(Idea courtesy Drew Sullivan)
--dave
Hmmn, I reserve that for posts whih have high scores, but contain factual errors credibly corrected in subsequent posts.
--dave
It's not sufficient to say "Y happened after X, therefor X caused Y". This is so old a logical error that it has a latin name (:-))
To prove X causes Y, you really have to show there is a causal realtionship, not just propinquity.
--dave
My former Director at Geac, Jacob Slonim, had a standard policy to keep people engaged, learning and growing in value to both themselves and the company: If you went for a promotion on the tech ladder, he'd second you to the business/management side for at least a quarter. A programmer got to learn what a business analyst does, an architect gets to learn team management, and so on.
Net result? Senior engineers with insight into the business, and sensitivity about not being "the unmanagable engineer" (;-))
--dave
--dave
An example I know about from an old version of Samba is that using branch prediction to compile a binary that expects all debugging code to be branched-around shows no detectable speed improvement.
It turned out that a complex debug macro expanded to a chunk of code which was bigger than a cache line on the machine I had. Net result? The branch, predicted or not, causes a cache refill, whose time completely dominates the time saved/spent by the branch prediction.
To use this optimixation, therefor, you would need to dedicate more silicon to cache line length in the i-cache and optimize the debug code for size, not speed. An interesting problem in hardware-software co-design (;-))
--dave
Hmmn, that was redundant, wasn't it (;-))
CISC also suffered from complex decoders, and required pipelines as long as two or three steps. This was considered scary-complex at the time.
I suspect a modern CISC might push similar complexity barriers...
--dave
See Kraft, Phillip, Programmers and Managers (Springer-Verlag) for details.
--dave
--dave
--dave
Average developers tend to notice when their compile speed drops by a factor of four, because they or their sysadmin replaced a 4-disk stripe with a single larger disk.
Even average users might notice if their programs and data files load four times slower...
--dave
The vendor's lawyer probably could have won that battle, but be dragged right back into court by the crown adding a copyright charge, and would then have lost. So it was better to cop a plea **before** they pissed off the crown attorney (the District Attorney in the US).
The important part is that the defence using the contract failed miserably,and should arguably fail the same way in the US.
In addition, one might usefully ask if the EULA, contract or other licence requires one to break the law, including inadvertantly joining a criminal conspiracy to keep bugs a secret.
--dave
--dave
I think you're replying to another's posting: I was accusing the antivirus company of conspiracy in restraint of trade (:-))
Some years ago, a CP/M modem program had a license which prohibited reverse engineering or disclosing anything about the program. Inspection proved that it was in fact a stolen copy of another modem program.
A purchaser reported this to the police, who promptly contacted the company that it had been stolen from, and prosecuted the thieves. They claimed the reverse engineering was illegal, but the court ruled that the contract was void, as it required the ultimate customer to break the law by keeping the theft a secret. It was therefor null and void.
If the anti-virus company was requiring the users to hide the fact the program was "not suitable for the purpose sold", then they arguably were demanding the users to enter into a criminal conspiracy in restraint of trade.
Users of such programs may wish to speak to their solicitors, lest by obeying the contract terms they inadvertently commit an offense (;-))
--dave
Instead, use Win4Lin or VMware to run just the Windows app you need. Which is, topically, my tax program (:-)) --dave
Each time they made money for the sellers of the scheme, but harmed the purchasers. And I don't mean the end-users, I mean the companies that shipped software that depended on unreliable and sometimes deliberately broken hardware.
Customers couldn't use the products, and returned them for a refund. Which made the dealers relctant to stock them, and eventually the products were supplanted by their more functional competitors.
--dave
- communicate with the project managers
- communicate with each other in the same language
- recognize problems the PMs faced, and
- contribute to solving them.
As you might expect, we saw every one of the cited problems, including pusilanimous management in the next two years of work. We did better that most at surviving, them, though!
--dave
Not that the games didn't make a buck or two (:-))
--dave
War is peace.
Freedom is Slavery.
Ignorance is Strength.
See Orwell, George, Nineteen Eighty-Four.
Both sound like lego bricks shoved into slots in a backplane (;-))
--dave