OK, so this afternoon I did an SQL test. I warned them that I haven't 
done any SQL for about 5 years and that I'm probably a bit rusty. OTOH, 
I was expecting it to be, like, *hard*? All they asked for was some 
pretty basic stuff like "find all pages that have been visited by 
Googlebot more than 5 times" and so forth. Pretty trivial SELECT statements.

[One required a union, several required grouping and aggregates. And I 
got tripped over by all the ones that want fuzzy text matching - I don't 
remember all that stuff any more.]

The last three completely stumped me. One wanted me to take some data 
from one table and insert it into another one. I know how that's done, I 
just don't remember the command syntax off the top of my head. Another 
wanted me to use "cursors" - something I've never ever done before. I 
don't even know what one *is*.

The final one was quite an interesting problem though... I'm wondering 
if anybody has a solution.

Given a table containing all the page hits, count how many visits the 
website has had. (Two hits are the same "visit" if they're the same 
username and happen within 10 minutes of each other.)

I sat for ages puzzling over this one. I can't figure out how the heck 
to do it... any pointers?

PS. Apparently "I see plenty of people who claim to know SQL who do 
worse than you did. You're really underselling yourself, Andrew. I 
wasn't even expecting you to ATTEMPT those last few..."

PPS. First person who says 0x3A28213A gets a slap!

PPPS. http://www.xkcd.com/138/

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

Orchid XP v7 escribió:
> any pointers?

NULL.

Post a reply to this message

Orchid XP v7 escribió:
> PS. Apparently "I see plenty of people who claim to know SQL who do 
> worse than you did. You're really underselling yourself, Andrew. I 
> wasn't even expecting you to ATTEMPT those last few..."

Remember that "unskilled and unaware of it" paper? The more you know, 
the more you know you don't know too (does that make any sense?). People 
may call themselves experts and not know what a JOIN is, and you say 
you're somewhat rusty because you know about lots of features that exist 
but you don't know very well, when you're actually much better than the 
idiot calling himself an expert.

Does that make any sense either? Seems it's impossible to explain this 
without long tricky sentences :)

Post a reply to this message

On Fri, 07 Mar 2008 16:58:58 +0000, Orchid XP v7 wrote:

> First person who says 0x3A28213A gets a slap!

LOL, you've got to know it's 0x3A28213B! ;)

Jim

Post a reply to this message

Nicolas Alvarez wrote:
> Orchid XP v7 escribió:
>> any pointers?
> 
> NULL.

If you give a person a NULL pointer, does that kill them?

Post a reply to this message

Jim Henderson wrote:
> On Fri, 07 Mar 2008 16:58:58 +0000, Orchid XP v7 wrote:
> 
>> First person who says 0x3A28213A gets a slap!
> 
> LOL, you've got to know it's 0x3A28213B! ;)

Neither, The alignment's all wrong on a 32-bit machine... it should be 
0x3A28213C

Post a reply to this message

Mike Raiford escribió:
> Jim Henderson wrote:
>> On Fri, 07 Mar 2008 16:58:58 +0000, Orchid XP v7 wrote:
>>
>>> First person who says 0x3A28213A gets a slap!
>>
>> LOL, you've got to know it's 0x3A28213B! ;)
> 
> Neither, The alignment's all wrong on a 32-bit machine... it should be 
> 0x3A28213C

Andrew, I think your warning not to give "that kind of pointers" has 
backfired badly.

Post a reply to this message

On Fri, 07 Mar 2008 11:17:03 -0600, Mike Raiford wrote:

> Jim Henderson wrote:
>> On Fri, 07 Mar 2008 16:58:58 +0000, Orchid XP v7 wrote:
>> 
>>> First person who says 0x3A28213A gets a slap!
>> 
>> LOL, you've got to know it's 0x3A28213B! ;)
> 
> Neither, The alignment's all wrong on a 32-bit machine... it should be
> 0x3A28213C

Oh, right, I forgot to carry the 1. ;-)

Jim

Post a reply to this message

Nicolas Alvarez wrote:

> Andrew, I think your warning not to give "that kind of pointers" has 
> backfired badly.

...apparently. *rolls eyes*

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

Nicolas Alvarez wrote:
> Orchid XP v7 escribió:
>> PS. Apparently "I see plenty of people who claim to know SQL who do 
>> worse than you did. You're really underselling yourself, Andrew. I 
>> wasn't even expecting you to ATTEMPT those last few..."
> 
> Remember that "unskilled and unaware of it" paper? The more you know, 
> the more you know you don't know too (does that make any sense?). People 
> may call themselves experts and not know what a JOIN is, and you say 
> you're somewhat rusty because you know about lots of features that exist 
> but you don't know very well, when you're actually much better than the 
> idiot calling himself an expert.
> 
> Does that make any sense either? Seems it's impossible to explain this 
> without long tricky sentences :)


Knowing what thou knowest not...
In in a sense...
Omniscience

Post a reply to this message

Orchid XP v7 wrote:
> just don't remember the command syntax off the top of my head. 

Maybe that's because different engines use different syntax. :-)

> Another 
> wanted me to use "cursors" - something I've never ever done before. I 
> don't even know what one *is*.

It's kind of like imperative Haskell - try to avoid it. ;-)

> I sat for ages puzzling over this one. I can't figure out how the heck 
> to do it... any pointers?

I would first find the duration of each page view, then look for page 
views more than 10 minutes apart.

The duration is the difference between the view time of this row and the 
min of the view time of all rows with a later view time (select by user 
name or whatever your criteria are).  Unobvious. I had to ask that one 
myself.

After that, find the distinct user/duration>=10 pairs.

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

Nicolas Alvarez wrote:

> Remember that "unskilled and unaware of it" paper? The more you know, 
> the more you know you don't know too (does that make any sense?). People 
> may call themselves experts and not know what a JOIN is, and you say 
> you're somewhat rusty because you know about lots of features that exist 
> but you don't know very well, when you're actually much better than the 
> idiot calling himself an expert.
> 
> Does that make any sense either? Seems it's impossible to explain this 
> without long tricky sentences :)

Hmm... it has a flavour...

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

Darren New wrote:
> I would first find the duration of each page view, then look for page 
> views more than 10 minutes apart.

Or, to clarify. Count the number of sessions that *ended*.  Much easier 
than counting the number of sessions that started.  Finding the end of a 
session is finding the row where there's no row with a greater timestamp 
but within 10 minutes of this timestamp.

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

Darren New wrote:
> Darren New wrote:
>> I would first find the duration of each page view, then look for page 
>> views more than 10 minutes apart.
> 
> Or, to clarify. Count the number of sessions that *ended*.  Much easier 
> than counting the number of sessions that started.  Finding the end of a 
> session is finding the row where there's no row with a greater timestamp 
> but within 10 minutes of this timestamp.

Care to write an SQL statement? ;-)

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

"Orchid XP v7" <voi### [at] devnull> wrote in message
news:47d17435$1@news.povray.org...


> The last three completely stumped me. One wanted me to take some data
> from one table and insert it into another one. I know how that's done, I
> just don't remember the command syntax off the top of my head.

INSERT INTO SomeTable (<Column names>)
SELECT <Columns> FROM SomeOtherTable
WHERE <Any relevant conditions>

> Another
> wanted me to use "cursors" - something I've never ever done before. I
> don't even know what one *is*.

Slow and inefficient.

> PS. Apparently "I see plenty of people who claim to know SQL who do
> worse than you did. You're really underselling yourself, Andrew. I
> wasn't even expecting you to ATTEMPT those last few..."

Yup. I see that too. People claiming to have 4 years of SQL experience and
can't write a  basic select statement

Post a reply to this message

"Orchid XP v7" <voi### [at] devnull> wrote in message
news:47d19c72$1@news.povray.org...
> Darren New wrote:
> > Or, to clarify. Count the number of sessions that *ended*.  Much easier
> > than counting the number of sessions that started.  Finding the end of a
> > session is finding the row where there's no row with a greater timestamp
> > but within 10 minutes of this timestamp.
>
> Care to write an SQL statement? ;-)

Very rough and probably wrong. It's late and I'm not going to test this

SELECT Stuff FROM someTable a
WHERE not exists
 (SELECT 1 FROM SomeTable b where DATEDIFF ('mi',a.timestamp,b.timestamp)
between 0 and 10 AND a.UserID = b.UserID and a.HitID != b.HitID)

That should give all hits that don't have another hit with the same user ID
in the next 10 minutes.

It's not particuarly efficient, but I don't care right now.

Post a reply to this message

On Fri, 07 Mar 2008 16:58:58 +0000, Orchid XP v7 <voi### [at] devnull> wrote:

>The final one was quite an interesting problem though... I'm wondering 
>if anybody has a solution.
>
>Given a table containing all the page hits, count how many visits the 
>website has had. (Two hits are the same "visit" if they're the same 
>username and happen within 10 minutes of each other.)

In Oracle, assuming userid is the user and visittime is the timestamp, I'd do it like
this ...

<SQL>

select 
   userid,
   sum(case when (visittime-priortime) > 10/1440 then 1 end) visits
from (
   select 
      userid, 
      visittime,
      nvl(lag(visittime) over (partition by userid order by visittime), visittime-1)
priortime 
   from visittest
   )
group by userid;

</SQL>

I can tell you for sure that I wouldn't have been able to pull that our of my butt on
the fly while testing.  I would have had an idea of how to do it, but would't recall
the syntax.


>PPPS. http://www.xkcd.com/138/


heh.

Post a reply to this message

On Fri, 07 Mar 2008 18:04:53 +0000, Orchid XP v7 wrote:

> Nicolas Alvarez wrote:
> 
>> Andrew, I think your warning not to give "that kind of pointers" has
>> backfired badly.
> 
> ...apparently. *rolls eyes*

I think you just rolled "snake eyes"!  (I'm seeing two eyes with one pip 
each)

Jim

Post a reply to this message

> If you give a person a NULL pointer, does that kill them?



Only in those "fatal exceptions" :)

H²

Post a reply to this message

>> If you give a person a NULL pointer, does that kill them?
> 
> Only in those "fatal exceptions" :)

GAAAAH! >_<

Please stop... :-{

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

Gail Shaw wrote:

>> PS. Apparently "I see plenty of people who claim to know SQL who do
>> worse than you did. You're really underselling yourself, Andrew. I
>> wasn't even expecting you to ATTEMPT those last few..."
> 
> Yup. I see that too. People claiming to have 4 years of SQL experience and
> can't write a  basic select statement

Yeah, well, like I said, the test was pretty basic. I was expecting them 
to test me on something, like, hard...

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

"Orchid XP v7" <voi### [at] devnull> wrote in message
news:47d26114$1@news.povray.org...
> Gail Shaw wrote:
>
> >> PS. Apparently "I see plenty of people who claim to know SQL who do
> >> worse than you did. You're really underselling yourself, Andrew. I
> >> wasn't even expecting you to ATTEMPT those last few..."
> >
> > Yup. I see that too. People claiming to have 4 years of SQL experience
and
> > can't write a  basic select statement
>
> Yeah, well, like I said, the test was pretty basic. I was expecting them
> to test me on something, like, hard...

That last question wasn't exactly easy.

Still, a lot of the people I get to interview wouldn't do well on that test.

Post a reply to this message

>> Yeah, well, like I said, the test was pretty basic. I was expecting them
>> to test me on something, like, hard...
> 
> That last question wasn't exactly easy.
> 
> Still, a lot of the people I get to interview wouldn't do well on that test.

As I said, most of the questions were like "list all the posts that have 
more than 10 posts, sorted in date order". Really taxing...

[In case it isn't clear, you'd need to perform a join for that. But even 
so...]

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

"Orchid XP v7" <voi### [at] devnull> wrote in message
news:47d27e74@news.povray.org...
> >> Yeah, well, like I said, the test was pretty basic. I was expecting
them
> >> to test me on something, like, hard...
> >
> > That last question wasn't exactly easy.
> >
> > Still, a lot of the people I get to interview wouldn't do well on that
test.
>
> As I said, most of the questions were like "list all the posts that have
> more than 10 posts, sorted in date order". Really taxing...

And as I said, a large number of people I've interviewed would have done
really badly on that.

Do I have to repeat myself any more times? I did comprehend what you wrote
first time.

Post a reply to this message

Orchid XP v7 wrote:
> Yeah, well, like I said, the test was pretty basic. I was expecting them 
> to test me on something, like, hard...

Last time I was interviewing, I asked people

"Write a program that prints the prime numbers from 2 to N, where N is 
passed on the command line."

The first three "experienced programmers" we were hiring couldn't manage 
that with 30 minutes and hints.

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

Kyle wrote:
> I can tell you for sure that I wouldn't have been able to pull that our of my butt
on the fly while testing.  I would have had an idea of how to do it, but would't
recall the syntax.

I believe there's a way simpler way of doing it that doesn't use any 
trinary operators or server-specific features.  It's brutally 
inefficient, but that wasn't a criterion. :-)

I'll go pull it out of the code I wrote to solve exactly this problem, 
next time I'm in that partition.

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

Darren New wrote:

> Last time I was interviewing, I asked people
> 
> "Write a program that prints the prime numbers from 2 to N, where N is 
> passed on the command line."
> 
> The first three "experienced programmers" we were hiring couldn't manage 
> that with 30 minutes and hints.

Maybe they don't know what a "prime number" is? (A surprising number of 
people don't know this. And of the ones who know what one is, a 
surprising number of them can't figure out how to FIND such numbers, or 
even CHECK one...)

OOC, are prime numbers [or number theory generally] relevant to what 
you're hiring these people for?

OTOH, it may just be that they're not very good programmers. ;-)

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

Orchid XP v7 escribió:
> Maybe they don't know what a "prime number" is? (A surprising number of 
> people don't know this. And of the ones who know what one is, a 
> surprising number of them can't figure out how to FIND such numbers, or 
> even CHECK one...)

Isn't that elementary school stuff?

Post a reply to this message

Nicolas Alvarez wrote:
> Orchid XP v7 escribió:
>> Maybe they don't know what a "prime number" is? (A surprising number 
>> of people don't know this. And of the ones who know what one is, a 
>> surprising number of them can't figure out how to FIND such numbers, 
>> or even CHECK one...)
> 
> Isn't that elementary school stuff?

Depends.

I went through my entire school career without ever having heard of a 
"prime number". I learned that when I went to college. Not because it 
was on the syllabus, but because I was bored so I spent 2 years *living* 
in the college library...

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

Orchid XP v7 wrote:
> Darren New wrote:
> 
>> Last time I was interviewing, I asked people
>>
>> "Write a program that prints the prime numbers from 2 to N, where N is 
>> passed on the command line."
>>
>> The first three "experienced programmers" we were hiring couldn't 
>> manage that with 30 minutes and hints.
> 
> Maybe they don't know what a "prime number" is? 

That was the "with hints." They not only knew what a prime number was, I 
basically told them the algorithm. As in "find all the numbers between 2 
and N which, when you divide them by every other number, they don't come 
out even."

> OOC, are prime numbers [or number theory generally] relevant to what 
> you're hiring these people for?

It wasn't. It was just a simple "are you blowing smoke on your resume" 
test.  Kind of like the guys who say "print the numbers from 1 to 100, 
but substitute "foo" for every number divisible by 5, and "bar" for ever 
divisible by 7, and "zoop" for ever divisible by both".  Amazing the 
number of people who choke under the simplest pressure, even if they 
*do* know how to program.



Similarly, one place I worked, Jerry Hall was the system admin for unix 
machines. He interviewed about 40 people before they could answer his 
three "troll questions". And when he showed me, *I* knew the answers.

1) You have two machines on a network. One has a big disk, the other has 
a tape drive and small disk. You have a tar tape with too much 
information to fit on the small disk. How do you untar it onto the big disk?

2) In the ls-listing drwxrsxrwx what does the s mean?  (Note that if you 
said "look in the ls man page" or even what the name of the flag is, it 
was good enough, even if you didn't know the semantics the flag implied.)

I don't remember what the third one was, but it was pretty much the same 
level. The sort of thing which, if you don't know how to solve it, you 
certainly shouldn't know the root password.

I also have a list of "interview questions" for testing the breadth of 
someone's knowledge. If I was ever hiring someone as a primary 
programmer for a new and growing company, I'd want them to know more 
than just Web2.0 javascripting. :-)

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

> 1) You have two machines on a network. One has a big disk, the other has 
> a tape drive and small disk. You have a tar tape with too much 
> information to fit on the small disk. How do you untar it onto the big 
> disk?

OK, I have no clue how to do that.

But then, I don't know how to do almost anything network-related in Unix 
beyond configuring the IP address and default gateway.

> 2) In the ls-listing drwxrsxrwx what does the s mean?  (Note that if you 
> said "look in the ls man page" or even what the name of the flag is, it 
> was good enough, even if you didn't know the semantics the flag implied.)

Hmm. Isn't that SetUID or Sticky or something?

> I don't remember what the third one was, but it was pretty much the same 
> level. The sort of thing which, if you don't know how to solve it, you 
> certainly shouldn't know the root password.

And that's the difference: I wouldn't be applying for a Unix sysadmin 
job. ;-) [Or at least, not unless they provide nontrivial training first...]

> I also have a list of "interview questions" for testing the breadth of 
> someone's knowledge. If I was ever hiring someone as a primary 
> programmer for a new and growing company, I'd want them to know more 
> than just Web2.0 javascripting. :-)

I'm still fuzzy on exactly what the hell "Web 2.0" actually is. Does 
anybody actually have a definition for it?

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

On Sat, 08 Mar 2008 20:55:28 +0100, Orchid XP v7 <voi### [at] devnull> wrote:
> I'm still fuzzy on exactly what the hell "Web 2.0" actually is.

It is a fuzzy term.


> Does anybody actually have a definition for it?

Sort of:

http://en.wikipedia.org/wiki/Web_2.0


-- 
FE

Post a reply to this message

Fredrik Eriksson wrote:
> On Sat, 08 Mar 2008 20:55:28 +0100, Orchid XP v7 <voi### [at] devnull> wrote:
>> I'm still fuzzy on exactly what the hell "Web 2.0" actually is.
> 
> It is a fuzzy term.

That's more or less what I thought. (Or I'd have shut up by now...)

>> Does anybody actually have a definition for it?
> 
> Sort of:
> 
> http://en.wikipedia.org/wiki/Web_2.0

Riiiight...

So it's like Web 1.0, but more enterprisy? :-D

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

Orchid XP v7 wrote:
> Hmm. Isn't that SetUID or Sticky or something?

That's better than many of the people he interviewed to be a sysadmin.

> I'm still fuzzy on exactly what the hell "Web 2.0" actually is. Does 
> anybody actually have a definition for it?

You know what "B2B" and "B2C" means? Business to Business and Business 
to Consumer? I think most of the Web 2.0 stuff is "Consumer to Consumer".

Of course, it started off as a marketing term. O'Reilly said "Let's call 
it Web 2.0. What will it be about?"

There are other terms you should be familiar with too:

"Developer's Wiki" - We're too lazy to actually write documentation on 
how to use the system, so we let the users reverse-engineer it and 
maintain the documentation for us. This has the additional advantage of 
having all documentation inaccessible whenever our server is down, which 
is often, because we don't care enough to write the documentation in the 
first place. And *another* advantage that there's no good starting 
point, and no way to know when the feature you're looking for *isn't* 
available!

"XML" - We're too lazy to document our data interchange formats, so 
we're hoping that by using an inefficient format, it'll include enough 
information that you can guess what we mean by looking at examples. 
Plus, it lets us pretend to parse it correctly by writing simple but 
incorrect code ourselves!

"REST" - Rather than use an RPC format that is capable of being 
generated automatically like SOAP, which (like XML) we don't really 
understand, we're going to use a catchy phrase like "REST" which is 
completely inapplicable in all practical implementations. Plus, it gives 
us the advantage of making you reimplement all parsing and security 
mechanisms for every service with which you interact.

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

Orchid XP v7 <voi### [at] devnull> wrote:
> Fredrik Eriksson wrote:
> > On Sat, 08 Mar 2008 20:55:28 +0100, Orchid XP v7 <voi### [at] devnull> wrote:
> >> I'm still fuzzy on exactly what the hell "Web 2.0" actually is.
> > 
> > It is a fuzzy term.

> That's more or less what I thought. (Or I'd have shut up by now...)

  "Web 2.0" is the conglomeration of the WWW technologies which have been
developed over the years.

  "Web 1.0" is: You request a webpage, you get a webpage (with fixed
contents). That's it.

  Gmail is a good example of "Web 2.0": The contents of the page are
updated on request, from dynamic information in the server side. Also,
you can send information (such as, for example, a WIP email) to the server
and it can store it and show it to you later. All this without having to
reload pages or jump between different pages.

  Many people have the opinion that "Web 2.0" is just a hype word.
It really isn't. It really has some concrete ideas behind it.

> So it's like Web 1.0, but more enterprisy? :-D

  No, it's Web 1.0 with the various dynamic interactivity technologies
which have appeared in the WWW during the years.

-- 
                                                          - Warp

Post a reply to this message

Darren New wrote:
> Orchid XP v7 wrote:
>> Hmm. Isn't that SetUID or Sticky or something?
> 
> That's better than many of the people he interviewed to be a sysadmin.

Woohoo! I'm better than somebody! :-D

Oh, wait...

> You know what "B2B" and "B2C" means? Business to Business and Business 
> to Consumer? I think most of the Web 2.0 stuff is "Consumer to Consumer".
> 
> Of course, it started off as a marketing term. O'Reilly said "Let's call 
> it Web 2.0. What will it be about?"
> 
> There are other terms you should be familiar with too:
> 
> "Developer's Wiki" - We're too lazy to actually write documentation on 
> how to use the system, so we let the users reverse-engineer it and 
> maintain the documentation for us. This has the additional advantage of 
> having all documentation inaccessible whenever our server is down, which 
> is often, because we don't care enough to write the documentation in the 
> first place. And *another* advantage that there's no good starting 
> point, and no way to know when the feature you're looking for *isn't* 
> available!
> 
> "XML" - We're too lazy to document our data interchange formats, so 
> we're hoping that by using an inefficient format, it'll include enough 
> information that you can guess what we mean by looking at examples. 
> Plus, it lets us pretend to parse it correctly by writing simple but 
> incorrect code ourselves!
> 
> "REST" - Rather than use an RPC format that is capable of being 
> generated automatically like SOAP, which (like XML) we don't really 
> understand, we're going to use a catchy phrase like "REST" which is 
> completely inapplicable in all practical implementations. Plus, it gives 
> us the advantage of making you reimplement all parsing and security 
> mechanisms for every service with which you interact.

Woah - suddenly I don't think it's *me* who should be starting their own 
web comic... ;-)

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

On Sat, 08 Mar 2008 11:32:28 -0800, Darren New <dne### [at] sanrrcom> wrote:

>I believe there's a way simpler way of doing it that doesn't use any 
>trinary operators or server-specific features.  It's brutally 
>inefficient, but that wasn't a criterion. :-)
>
>I'll go pull it out of the code I wrote to solve exactly this problem, 
>next time I'm in that partition.

I imagine there are several ways to skin that cat.  I'm sure I can write another
version by joining
a couple of subqueries on the table, but it would likely be quite inefficient, as you
stated.  I
don't know about it being simpler, though.

Post a reply to this message

Kyle wrote:
> I imagine there are several ways to skin that cat.

Simpler conceptually:

It was something like this. We have a table of data coming in,
different sources providing us with data, and we know the
starttime, but we have to also know how long the data was
valid.  So, roughly, and with the wrong syntax:

create table metadata (
   starttime, duration default null,
   source, information
)

update metadata as y set duration =
   (select min(starttime) from metadata as x
     where y.source = x.source and
           y.starttime < x.starttime)
   where y.duration is null

After that sort of thing ran overnight without finishing,
I said screw it, and wrote a program to go through
the table and do the update programmatically, which
ran in about five minutes or so. With the ability
to tell how far you were finished.

Then, to answer the question Andrew actually was asked, you'd modify it 
to look for or count durations > 10 minutes, rather than storing them in 
the table.  That's what I meant about finding the ends of sessions 
instead of the start.

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

And lo on Sat, 08 Mar 2008 19:51:00 -0000, Darren New <dne### [at] sanrrcom>  
did spake, saying:

> 1) You have two machines on a network. One has a big disk, the other has  
> a tape drive and small disk. You have a tar tape with too much  
> information to fit on the small disk. How do you untar it onto the big  
> disk?

I'm confused - can't the big disk machine access the tape drive directly  
over the network?

-- 
Phil Cook

--
I once tried to be apathetic, but I just couldn't be bothered
http://flipc.blogspot.com

Post a reply to this message

Phil Cook wrote:
> I'm confused - can't the big disk machine access the tape drive directly 
> over the network?

No. How would you do that?

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

And lo on Tue, 11 Mar 2008 19:53:58 -0000, Darren New <dne### [at] sanrrcom>  
did spake, saying:

> Phil Cook wrote:
>> I'm confused - can't the big disk machine access the tape drive  
>> directly over the network?
>
> No. How would you do that?

As part of a SAN

-- 
Phil Cook

--
I once tried to be apathetic, but I just couldn't be bothered
http://flipc.blogspot.com

Post a reply to this message

On Wed, 12 Mar 2008 09:04:29 +0000, Phil Cook wrote:

> And lo on Tue, 11 Mar 2008 19:53:58 -0000, Darren New <dne### [at] sanrrcom>
> did spake, saying:
> 
>> Phil Cook wrote:
>>> I'm confused - can't the big disk machine access the tape drive
>>> directly over the network?
>>
>> No. How would you do that?
> 
> As part of a SAN

Well, if the tape device was, that would be one way.  I didn't assume 
that when I thought of my answer, which would be just to export the big 
disk machine's filesystem to the small disk machine using NFS.

Jim

Post a reply to this message

Jim Henderson wrote:
> that when I thought of my answer, which would be just to export the big 
> disk machine's filesystem to the small disk machine using NFS.

I guess that would work. For years I'd been using

cat /dev/tape | rsh overthere tar xf -



-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

On Wed, 12 Mar 2008 20:18:17 -0700, Darren New wrote:

> Jim Henderson wrote:
>> that when I thought of my answer, which would be just to export the big
>> disk machine's filesystem to the small disk machine using NFS.
> 
> I guess that would work. For years I'd been using
> 
> cat /dev/tape | rsh overthere tar xf -

That'd work, too - obviously. ;-)

Jim

Post a reply to this message