Re:I turned down a well paying job at Walgreens
on
Suit Up Or Ship Out?
·
· Score: 5, Insightful
Additionally, they worked wierd for IT hours, of only 8:00-4:30. They do not work overtime, weekends,or anything else. I didn't
want to be in a programming department that was that regimented. It is a creative process, and if I wanted to work late to figure out
a problem, they didn't want that.
Actually, the most challenging software
engineering jobs I know of are purely "9 to 5"
(or whatever regular hours) jobs. These are CMM level 5
shops, and work on little simple programs
like the Space Shuttle guidance and control
software.
That's not to say that "wear a suit" is a
requirement at those shops, but the idea is that
leadership and cohesiveness are vastly important
to reliable software. In other words, the
space shuttle isn't going up guided by code
that a guy wrote late last night:-).
I think they are right. We as mankind have already destroyed more forests than we need.
but agreement as to what the policy should or
shouldn't be isn't my issue. My issue is that
the US, a country of rich fat people, is trying
to tell poor starving people in foreign countries
what they cannot do to improve their lives.
The example you give of economic incentives
to do the "right thing" is the best route, but
I don't see this happening as much as it should.
It doesn't seem to be mentioned much in the
mainstream press, but here in the DC area
AOL is aggressively hiring software engineers
with Linux/Perl/CGI/database experience for
their "internal" functions. One would suppose
that this will reduce the cost and increase the
efficiency of their back-office functions, after
they fire all those MSCE's that run around doing
the retry/reboot/reinstall cycle on their current
internal network of MS machines.
Most likely they'd refer you to the local
mental health provider. Random weirdos do show up and demand to be
prosecuted for random things, some of which
are "real crimes", some of which aren't. Almost
certainly you'd fall into what the local police
think are not "real crimes".
Basically the "staff" appear to wonder if it wouldn't
have been more productive to give the $170,000 that the League appears to have collected (or be owed)
directly to the KDE project."
But then the League wouldn't have that
cool metallic headquarters downtown headquarters
from which they fight crime.
It's the natural color of the epoxy resin usually used in PCB's based on G-10 glass epoxy.
Masking to get other colors is generally pretty
cheap. In some cases (for example UV resistance)
dyes can be put in the glass epoxy at manufacture
time.
My home web server, through a UUnet SDSL connection,
has about a third the usual traffic. Problems
started about 5AM this morning, when the whole
outside world would disappear for a couple minutes
at a time. Peering connections with other big
ISP's here in the DC area seem to have been up
and down all day, and in all there's about
a third the traffic I'd normally expect on a
weekday.
Here at work (not served directly by UUNet)
service to various websites has been intermittently down for up to a few hours at a
time.
Re:Upgraded yesterday, Apache migration info
on
Red Hat 8.0 Released
·
· Score: 2
Interesting experience you posted there; I appreciate the info.
But it mostly confirms what I've thought all along: RedHat shouldn't be supplying any
"default" Apache installs, just as Windows
shouldn't be supplying any "default" IIS installs.
Everyone who sets up a server, IMHO, should
build the components from source and test them
out on not-port-80. And, as an aside to the
way Redhat clobbered your old Apache install, if you had
built Apache 2.x from sources it would have
(by default) installed itself in/usr/local/apache2 to keep it
distinct from/usr/local/apache,
the default place to plop Apache 1.x.
Assume you want to supply everyone with DSL equivelent speeds - 40 kByte/sec....
I don't think any commercial broadband wired
services would be viable if everyone used all
their available bandwidth all the time. For
example in the past year or so most cable
companies have started putting download caps
into effect, for very good reason: you cannot
sell bandwidth that costs you, e.g. $600
a month for a T1 to consumers for $30 a month.
Never mind that the coax they use cannot supply
that much bandwidth to more than a few folks
per neighborhood.
A more realistic TCP/IP-by-satellite involves
intermittent (on-the-go) usage or more efficient multicast broadcasts. No, it's not a
T1-type tarrifed service anymore!
A Resume is a short (one or two page) attempt
to sell yourself to HR
A CV is a detailed description of your experience (notably published papers in academia, but in the software world it would likely list packages you wrote
or groups you worked in)
AOL has been looking for Linux engineers lately
on
AOL's new Linux PC
·
· Score: 5, Interesting
Here in the DC area AOL has been looking
for a large number of Linux software engineers
as of late. I always thought that these were
for "back-office" applications (account management
etc., heavy desire for Perl and database
experience) maybe some other positions seem to be
oriented towards end-user applications.
But clearly a few of the senior
coauthors appear to have been severely negligent, or criminally lazy.
Some fine institutions are so good at internally
refereeing their own papers that if it gets
submitted to Phys Rev, it's almost guaranteed
to be published.
Other institutions are not so good, and random
junk comes out.
I would guess that Bell Labs would like to be nearer
the "fine institutions" rather than the junk
ones.
The (external) referees that approved the papers
for publication deserve some of the blame too, but
not all of it - when it comes to the first data
of its kind, there really isn't much they can
compare against.
As a non-scientist, I have a pretty difficult time understanding what the "difficult issue" is; if an author's not ready to stand behind a
paper published under his name, what's his name doing on it?
There's incredible pressure to publish, even
if your name isn't first.
The convention as to whose name goes on the
paper and whose name goes first varies throughout
academia. In Biology, for example, it is very
common for the guy who got the funding to be
named first, even if he didn't do any of the
work. In Physics it's a bit more equitable
most of the time, but not always.
Large collaborations (often there are hundreds
of authors in a big collider experiment) have
committees to decide on what's published and
what's not. In some cases your name automatically
goes on all collaboration publications unless
you specifically object.
I'm not sure; my guess was that this was
incompetence caused by HP's managers.
To get folks up to speed, HP blamed low
benchmark scores on one HP engineer. Then
they fired him. Then they sued him. Makes
you think twice about recommending any set of
compiler flags, doesn't it?:-(
See this Register article.
The Bell Labs case is different; it's much
more a case of integrity rather than just
benchmark scores. Clearly the labs felt that
their integrity was being hurt by one sore
thumb, but I do not see it at all as a bunch
of vindictive uppity-up's taking their wrath
out on a little guy.
Why would you not want to encrypt everything is beyond me...
I certainly use it for passwords and anything
with any possible financial impact. But I
don't see the purpose of doing it for much else.
Maybe it's just a habit I picked up from reading
all those crypto books in grade school, but it's
well known that the greater the number of
intercepts, the easier it'll be for someone to
crack a code. Not that I believe those numbers
are anything but zero for 128-bit encryption:-)
The problem is that many (most? all?) the big-name
distros have Apache built with mod_ssl on them. Even
though I would guess that only a tiny percent
of all web servers need SSL. (Admittedly that
tiny percent is very important, as no money
transactions should be going on without security...)
IMHO if you need SSL on a webserver, you should
be forced to go through the download + build +
cert process yourself.
Microsoft *is* the choice for Dept of Interior
on
USDOI Goes 100% Microsoft
·
· Score: 5, Insightful
What the head article fails to mention is that
a Federal judge ordered the Department of the
Interior to shut down all internet connections last year. With
no from-the-outside network attacks, the
Microsoft systems might stay up for days, even.
The sites referenced in this article make it
sound like this is the first and only way
for small web-sites to get payments. This
clearly isn't the case; Paypal has
been delivering this functionality for a couple
of years now, and even Amazon's earlier
zShops worked for a lot of
small-time web stores.
What is important to point out is that not only
are the
technology and methods of all these approaches
different, but the legal standards and rules
of conduct all vary as well. For example,
lots of folks despise Paypal and Amazon's zShops
never really took off, in part because they
closely regulate sellers (but for other reasons
related to Amazon's fundamental business model too.)
We don't need to work on the spindle speed. They're working on data density instead. Think about it. The data density on these drives is 4 times
that of an 80GB drive. So if the data transfer on the 80GB drive is X*7200rpm, the 320GB is 4X*5400rpm = 3 times the raw throughput.
But worse rotational latency. That's the point
of high-RPM drives, after all.
Yeah, but you don't get the MB/sec transfer rate, so what's the point?
Faster seeks! Reduce the rotational latency
by spinning the platter faster and you'll have
to wait less time for the data to come under
the head.
If you do streaming video, seek times may not
matter much to you. But for many applications
which have large numbers of small files, seek
times are usually the limiting factor. There's
much more than just MB/s when it comes to disk
performance.
And it is easier to put 48 SCSI drives into a PC-clone case?
No, but you can put 48 (or 480, or 4800) SCSI
drives outside the PC-clone case. This isn't
an option with IDE, where cable-length
limitations hit you real fast.
I agree, no desktop user needs that many drives,
and few server platforms truly need that many either. But it's available for those who do.
Again, I'm no SCSI bigot; all my personal
systems are now ATA. But there is a very real
market segment where ATA is not
an option, either for RPM or drive number/cable
length reasons.
Trouble with SCSI is they keep upgrading the specs so
you have to get a new card if you want to go to the next higher speed.
Not spindle speed. You can put the newest
fastest 15K RPM SCSI drive on an 15-year-old
computer with a SCSI-1 bus. You probably need
a SCA to 50-pin Centronics chain of adapters,
and of course the drive will fall back to
single-ended mode as opposed to low-voltage differential,
but it works.
So in otherwords, you are a really good perl coder when you are the only one that can read your code
No, that's not what I said. I said that
qualities of good code include using "idomatic
Perl" (i.e. not writing C in Perl) and making
use of Perl's strengths (like hashes). Do these and you'll
be a good ways towards writing readable code.
Slashdot Mirror
Actually, the most challenging software engineering jobs I know of are purely "9 to 5" (or whatever regular hours) jobs. These are CMM level 5 shops, and work on little simple programs like the Space Shuttle guidance and control software.
That's not to say that "wear a suit" is a requirement at those shops, but the idea is that leadership and cohesiveness are vastly important to reliable software. In other words, the space shuttle isn't going up guided by code that a guy wrote late last night :-).
The example you give of economic incentives to do the "right thing" is the best route, but I don't see this happening as much as it should.
It doesn't seem to be mentioned much in the mainstream press, but here in the DC area AOL is aggressively hiring software engineers with Linux/Perl/CGI/database experience for their "internal" functions. One would suppose that this will reduce the cost and increase the efficiency of their back-office functions, after they fire all those MSCE's that run around doing the retry/reboot/reinstall cycle on their current internal network of MS machines.
Most likely they'd refer you to the local mental health provider. Random weirdos do show up and demand to be prosecuted for random things, some of which are "real crimes", some of which aren't. Almost certainly you'd fall into what the local police think are not "real crimes".
But then the League wouldn't have that cool metallic headquarters downtown headquarters from which they fight crime.
Oh, wait, that was the Justice League.
Masking to get other colors is generally pretty cheap. In some cases (for example UV resistance) dyes can be put in the glass epoxy at manufacture time.
Again, the most prominent, first-mentioned, feature of the Intel reference motherboard is its... Black Color.
Here at work (not served directly by UUNet) service to various websites has been intermittently down for up to a few hours at a time.
But it mostly confirms what I've thought all along: RedHat shouldn't be supplying any "default" Apache installs, just as Windows shouldn't be supplying any "default" IIS installs. Everyone who sets up a server, IMHO, should build the components from source and test them out on not-port-80. And, as an aside to the way Redhat clobbered your old Apache install, if you had built Apache 2.x from sources it would have (by default) installed itself in /usr/local/apache2 to keep it
distinct from /usr/local/apache,
the default place to plop Apache 1.x.
I don't think any commercial broadband wired services would be viable if everyone used all their available bandwidth all the time. For example in the past year or so most cable companies have started putting download caps into effect, for very good reason: you cannot sell bandwidth that costs you, e.g. $600 a month for a T1 to consumers for $30 a month. Never mind that the coax they use cannot supply that much bandwidth to more than a few folks per neighborhood.
A more realistic TCP/IP-by-satellite involves intermittent (on-the-go) usage or more efficient multicast broadcasts. No, it's not a T1-type tarrifed service anymore!
Here in the DC area AOL has been looking for a large number of Linux software engineers as of late. I always thought that these were for "back-office" applications (account management etc., heavy desire for Perl and database experience) maybe some other positions seem to be oriented towards end-user applications.
Some fine institutions are so good at internally refereeing their own papers that if it gets submitted to Phys Rev, it's almost guaranteed to be published.
Other institutions are not so good, and random junk comes out.
I would guess that Bell Labs would like to be nearer the "fine institutions" rather than the junk ones.
The (external) referees that approved the papers for publication deserve some of the blame too, but not all of it - when it comes to the first data of its kind, there really isn't much they can compare against.
There's incredible pressure to publish, even if your name isn't first.
The convention as to whose name goes on the paper and whose name goes first varies throughout academia. In Biology, for example, it is very common for the guy who got the funding to be named first, even if he didn't do any of the work. In Physics it's a bit more equitable most of the time, but not always.
Large collaborations (often there are hundreds of authors in a big collider experiment) have committees to decide on what's published and what's not. In some cases your name automatically goes on all collaboration publications unless you specifically object.
To get folks up to speed, HP blamed low benchmark scores on one HP engineer. Then they fired him. Then they sued him. Makes you think twice about recommending any set of compiler flags, doesn't it? :-(
See this Register article.
The Bell Labs case is different; it's much more a case of integrity rather than just benchmark scores. Clearly the labs felt that their integrity was being hurt by one sore thumb, but I do not see it at all as a bunch of vindictive uppity-up's taking their wrath out on a little guy.
I certainly use it for passwords and anything with any possible financial impact. But I don't see the purpose of doing it for much else.
Maybe it's just a habit I picked up from reading all those crypto books in grade school, but it's well known that the greater the number of intercepts, the easier it'll be for someone to crack a code. Not that I believe those numbers are anything but zero for 128-bit encryption :-)
IMHO if you need SSL on a webserver, you should be forced to go through the download + build + cert process yourself.
What the head article fails to mention is that a Federal judge ordered the Department of the Interior to shut down all internet connections last year. With no from-the-outside network attacks, the Microsoft systems might stay up for days, even.
What is important to point out is that not only are the technology and methods of all these approaches different, but the legal standards and rules of conduct all vary as well. For example, lots of folks despise Paypal and Amazon's zShops never really took off, in part because they closely regulate sellers (but for other reasons related to Amazon's fundamental business model too.)
But worse rotational latency. That's the point of high-RPM drives, after all.
Faster seeks! Reduce the rotational latency by spinning the platter faster and you'll have to wait less time for the data to come under the head.
If you do streaming video, seek times may not matter much to you. But for many applications which have large numbers of small files, seek times are usually the limiting factor. There's much more than just MB/s when it comes to disk performance.
No, but you can put 48 (or 480, or 4800) SCSI drives outside the PC-clone case. This isn't an option with IDE, where cable-length limitations hit you real fast.
I agree, no desktop user needs that many drives, and few server platforms truly need that many either. But it's available for those who do.
Again, I'm no SCSI bigot; all my personal systems are now ATA. But there is a very real market segment where ATA is not an option, either for RPM or drive number/cable length reasons.
Not spindle speed. You can put the newest fastest 15K RPM SCSI drive on an 15-year-old computer with a SCSI-1 bus. You probably need a SCA to 50-pin Centronics chain of adapters, and of course the drive will fall back to single-ended mode as opposed to low-voltage differential, but it works.
Then why are you buying IDE and not SCSI? 15K RPM is old-hat in the SCSI world.
If we wanted space, we'd just get additional drives.
Again, an area where SCSI shines. It's tough to put 48 IDE drives in a PC-clone case!
I'm not saying that SCSI is the solution for everyone, but it's been there and will continue to be there for the needs you mention.
No, that's not what I said. I said that qualities of good code include using "idomatic Perl" (i.e. not writing C in Perl) and making use of Perl's strengths (like hashes). Do these and you'll be a good ways towards writing readable code.