Re: Templating Solutions

2001-06-19 Thread Ian Brayshaw

snip type=inevitable love/hate circular debate/

I was going to stay quiet on this one (still don't know why I am now joining 
in).

I am finding XSLT  XML to be a good alternative to normal templating 
techniques. One of the biggest benifits I've found is being able to generate 
the one data set and have it rendered in different ways for different 
applications. I presume this is possible in TT2. H::T has the drawback of 
only allowing substitutions for tags defined in the template. Changing the 
template to render say a reduced set of data typically involves changing 
code.

I'm also free to choose my transformation platform, using something like 
XML::LibXML or Saxon on the server side, or just throwing it straight to the 
user and letting their browser take care of the rest.

Don't think DW jockeys will like the XSLT, but I'm fortunate in not having 
to deal with them.

My £0.02


Ian
(... trying desparately to avoid joining the XML bandwagon ...)
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




RE: *Buffy's Not Included

2001-06-14 Thread Ian Brayshaw

  From: Leo Lapworth [mailto:[EMAIL PROTECTED]]
 
  XML - do it because you need it, not because of the Buzz.

Amen.


I'm also trying to future proof the system slightly - I think that by 
having the data XML based it may make it easier to use in new and
wonderful ways in the future, without having to write all the tools from 
scratch.

I'm doing a lot of work at the moment with XML (still fairly rudiemntary 
stuff, though) and for me the future proofing has been the deciding 
factor. We don't know what we need the system to be capable of in the medium 
to long term, and so we've gone down the XML path to keep our options open.

The way I look at it, if it makes the data more readable and easily managed 
(through XPath  XSLT for example), then provided you don't suffer an 
unbearable performance hit, it's worth the effort.

Also, with something reasonably simple, it's probably a safe training ground 
for getting into XML without the pressure of a more critical project. Let 
you find out what you like/dislike about XML without becoming a lemming.


XML in a Nutshell is a very good book

++


My £0.02.


Ian

_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: Some pretty pictures ...

2001-06-11 Thread Ian Brayshaw

Piers Cawley [EMAIL PROTECTED] wrote:

Not very perlish I'm afraid.

a) I'm sure that will change in time (any camel shots?).
b) That makes it an appropriate topic for this list.


Ian

_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: M$ SQueaLServer

2001-06-07 Thread Ian Brayshaw

Hi guys,

Thanks for the input. I'll investigate further, but it has confirmed my 
suspicions that SQueaL hasn't made an impact at the terabyte level.

As for the job, it's good, but not the be-all and end-all. I'm too 
passionate about what I do to work on systems that I don't believe in. 
Anyway, it would be the excuse some are looking for to dump Perl/Apache and 
move to ASP/IIS/ColdFusion.

Might see you at the pub tonight and do a proper introduction.

Cheers.


Ian
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




M$ SQueaLServer

2001-06-06 Thread Ian Brayshaw

Hi guys,

Have any of you worked with SQueaLServer with a large DB (multiple terabyte 
level), serving high volume transactions (read  write, of the order of 
millions of records a day). What sort of performance did you get? What was 
the hardware? Was it reliable?

I'm working for a telecoms company that is considering a proposal to move 
its billing system from Oracle on Solaris, to SQueaLServer  NT. It's a 
decision that is coming from management (where else?), and I'm trying to 
find out if it's as ludicrous as it sounds.

My gut reaction is that it's still too warm in hell to consider this, but 
maybe I'm just showing my diehard support for *nix, and my desire to see the 
demise of the Evil Empire.

Any tales of first-hand experience (or old wive's tales) would be 
appreciated.


Ian
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: M$ SQueaLServer

2001-06-06 Thread Ian Brayshaw

Robin Szemeti [EMAIL PROTECTED] wrote:

On Wed, 06 Jun 2001, Ian Brayshaw wrote:

  I'm working for a telecoms company that is considering a proposal  to 
move its billing system from Oracle on Solaris, to SQueaLServer   NT. 
It's a decision that is coming from management (where  else?), and I'm 
trying to find out if it's as ludicrous as it sounds.

Ho ho .. Ian, April fools day was ages ago ...

I didn't even reallise you could get NT for serious mips .. I though it 
only ran on likkle PC things ...

I wouldn't have used the word ran ...

Yeah, I know this sounds crazy, but unfortunately the coding world is 
occasionally inhabited by the unwashed (or should that be brainwashed?). We 
have a strong (ie vocal) VB development team (sorry for swearing on this 
list; time to repent: pony pony pony buffy buffy buffy willow willow 
willow), who are advising how to proceed. We also have a new head of IT 
who is likely to support the move because it's the same system that I have 
on my desktop. No one seems to have drawn the connection with its place on 
the desktop and its inability to do anything beyond (and including) the 
desktop. The chief advisor raves about the power, flexibility and price of 
SQueaLServer19100 being more than a match for Oracle 8i/9i.

What I'm trying to find is industry evidence of SQueaL's performance (or 
lack of). The more gory the details the better. Our VB guru exclaims the 
ease with which a major New Zealand bank rolled out SQueaL on (what I can 
presume to be a truck load of) NT servers without a hitch. He's a nice 
guy, but he's living in La-la-land if he thinks the throughput of a Kiwi 
bank matches that of an international telco.

So far the wailing and gnashing of teeth by the *nix  DBA people have been 
ignored. Has anyone else had to deal with this sort of mind set? Any advice 
(apart from becoming a US postal worker...)

If it goes through, this is one coder that will be seeking alternate 
employment (along with the rest of the company).


Ian
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: M$ SQueaLServer

2001-06-06 Thread Ian Brayshaw

Leon Brocard [EMAIL PROTECTED] wrote:

Redvers Davies sent the following bits through the ether:

  The transactions world record sadly is held by M$ at the moment.

http://www.tpc.org/tpch/results/h-ttperf.idc

Yeah, seen that. It's interesting to note that SQueaL doesn't make an 
appearance at the terabyte level, which is what we're dealing with.

Ian


... Hmm... How *did* they finally kill Frosty?

Global warming?

_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: pc components

2001-05-17 Thread Ian Brayshaw

will [EMAIL PROTECTED] wrote something that looked like the following:

Apparently it is Craig 'Red Dwarf' Charles's regular and he was
there a few weeks ago when we were there. Not as pretty as Buffy or
Willow about as close as you can get (sort of).

I saw Craig Charles at the Melbourne Comedy Festival a couple of years ago 
and it was a waste of time and money. He walked out on stage, said he was 
p1ssed, drank beer in front of the audience for an hour, occasionally 
screamed juvenile jokes centred around his manhood, then suddenly declared 
that he'd had enough and swaggered off stage. Complete w4nker, really, given 
the price of the tickets.


Ian
(who'd like to meet either Buffy or Willow in a dark alley...)


_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: require Module; and filehandles

2001-04-27 Thread Ian Brayshaw

On Fri, Apr 27, 2001 at 01:32:24PM +0200, Philip Newton wrote:
  Ian Brayshaw wrote:
   it's the internal workings of require that stop the tie
   from being honoured. I presume that the require burrows
   down into the internals and isn't aware that it's a tie'd
   handle. As far as I can tell the code within the require
   call is unaware that this handle is an object. If you place
   an AUTOLOAD method in TrueHandle instead of the READ and
   READLINE methods, only DESTROY is called.
 
  Did you tell p5p about this? Perhaps they can do something about
  it, if they consider this a bug.

Having looked at the implementation yesterday, I pity the person
who's *so* certain it's a bug that they try and fix it. It would
not be an easy thing to do, I don't think. (Anyway, PerlIO layers
will save us all!)

The report has been sent. Will see what comes of it. I suspect nothing, 
given the new PerlIO layers.

Thanks for your help.


Ian
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: require Module; and filehandles

2001-04-26 Thread Ian Brayshaw

From: Jonathan Stowe [EMAIL PROTECTED]

So whats wrong with StatInc ?

OK, so StatINC will reload the module if the module source has changed. 
However, it's the handler I am pushing onto @INC that will generate the 
module source from a configuration file. If the configuration file changes, 
then the module needs to be regenerated and reloaded. I can't rely on a scan 
of %INC because the trigger for reloading the module is not the change of 
the module source, but the change in a secondary source.

Yes, one alternative would be to have a program monitor the configuration 
files and regenerate the modules whenever the configuration files change, 
but I'd prefer to have the regeneration occur when/if it is needed.

May be this will clarify what I'm trying to do (or let someone see where I 
have missed the boat):

sub generate_modules { .. }

use lib \generate_modules;

...

require Module;

Module.pm doesn't exist, but Module.conf does, and generate_modules() 
converts Module.conf into Module.pm which is then loaded.

Later, while the program is still running, Module.conf changes, and I'd like 
a further call to

require Module;

(after the change in Module.conf) to regenerate and reload Module.pm.

By removing an entry in %INC I can get require to call generate_modules() 
again, even though Module has already been loaded.

My original question was concerned with trying to stop the unnecessary 
recompilation of Module.pm that results when its entry in %INC is removed 
and

require Module;

is called, but there has been no change to the configuration file, and hence 
no need to regenerate or reload the .pm file.

I am trying to implement transparent module creation so that in the program 
I don't have to know about the configuration files, and can treat them as 
normal modules.

If none of this makes any clear sense then I'll disappear and think some 
more.

Thanks for your suggestions and patience.



Ian

(who should have paid more attention during his formative years to learn how 
people communicate effectively)
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: require Module; and filehandles

2001-04-26 Thread Ian Brayshaw

Robin Houston wrote:

snip/

In earlier perls, you'll need to make a tied filehandle:

Thanks Robin. The TrueHandle package is essentially what I had implemented, 
but it's the internal workings of require that stop the tie from being 
honoured. I presume that the require burrows down into the internals and 
isn't aware that it's a tie'd handle. As far as I can tell the code within 
the require call is unaware that this handle is an object. If you place an 
AUTOLOAD method in TrueHandle instead of the READ and READLINE methods, only 
DESTROY is called.


Ian

_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




require Module; and filehandles

2001-04-26 Thread Ian Brayshaw

Hi,

Sorry to pollute this list with a question about Perl...

I am writing a customer handler for loading modules at runtime, taking 
advantage of the support for coderefs in the @INC array. By deleting entries 
in the %INC hash for loaded modules I can force Perl to recompile the module 
after it has been first loaded. This is particularly important with 
mod_perl, since the modules themselves may change during the life of the 
program.

Deleting an entry in %INC allows me to trap the request from a 'require' 
call to load a module, even if it has already been loaded. If the module 
should be recompiled, then I simply return a filehandle to the module 
source. Sometimes, however, no recompilation is necessary, and so I'd like 
to return a filehandle that evaluates to true (in the 'do file' sense of 
evaluates), so that require thinks that it has successfully loaded and 
recompiled the module (to avoid the overhead of unneccessarily recompiling 
large modules).

Is there a way to do this without creating a dummy file (i.e. can we do this 
in memory)?

I have tried using a tie'd filehandle, but the internal workings of require 
mean that this doesn't work (unless, of course, I got my tie'ing wrong) as 
none of the tie'd package methods are called.


Cheers,


Ian

_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: The Natives are Revolting

2001-04-19 Thread Ian Brayshaw

Dave Cantrell wrote:

Hmmm ... if 'not working at all' is the same as 'not bad'.  All I
get is the title image.

Got to love badly formed HTML (probably a missing /table tag or two). ... 
And I thought he was an "master". ... Oops, sorry, that was only in "PERL", 
not HTML.

(... Mind you, it all looks lovely(?) and dandy(?) under IE ... you've got 
to love the "near enough is good enough" approach of M$ ...)


Ian

_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




RE: Tech mtg?

2001-04-18 Thread Ian Brayshaw

http://www.streetmap.co.uk/streetmap.dll?G2M?X=533817Y=182479A=YZ=1

Thanks Jo.


Ian

_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: Contract market these days?

2001-03-23 Thread Ian Brayshaw

Just wondering what it was like in the UK right now.

snip

... what the overall feel of the market is right now whether people
are knifing coming in and out of interviews or if it's pretty easy
to find IT work, etc.

It seems to be quite strong, but definitely not as buoyant as it was. A 
number of traditional sources of contract work seem to be drying up (a few 
banks are sacking a lot of contractors in favour of permanents), but at the 
same time there still a steady stream of people needing the temporary work 
force. The dot bomb market has kind of leveled off and there's not as much 
money being thrown around, but the market is a long way from being on 
death's door.

If you've got the skills there's plenty of work (as always).


My £0.02.


Ian
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: That book

2001-03-23 Thread Ian Brayshaw

Dave Cross wrote:
At Fri, 23 Mar 2001 16:37:51 + (GMT), Mark Fowler [EMAIL PROTECTED] 
wrote:
  L. said:
 
   World domination is ours.  Muahahaha!
 
  Not if we can't come up with a good name for not matt's script
  archive it won't be.
 
  Ideas to the usual address.
 
  *Please*

I thought that 'EasyScripts' (or, even, 'EZScripts') had most
people's approval. I thought it ws alright - well, taking the target
audience into account :)

Dave...

What about "Web Scripts" or "Web Tools"? Maybe it's just me, but "EZScripts" 
sounds like the LCD, and while these people are the target, the archive 
itself is not the LCD. (Maybe I just miss the point of targeted marketing.)


Ian
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




lvalue subroutines

2001-02-27 Thread Ian Brayshaw

Hi,

Given the following lvalue subroutine

sub mysub : lvalue {
$value;
}

is there any way for mysub() to be able to determine that it
was called in an lvalue context?


Ian (New face on london-pm: usually wears an over-the-top suit.)
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: lvalue subroutines

2001-02-27 Thread Ian Brayshaw

Ian Brayshaw" [EMAIL PROTECTED] writes:

  Hi,
 
  Given the following lvalue subroutine
 
  sub mysub : lvalue {
  $value;
  }
 
  is there any way for mysub() to be able to determine that it
  was called in an lvalue context?

No. If you need to know that sort of thing, you kind of have to tie
the $value that you're going to return, and use that as a proxy for
the *actual* value. If you're called in an lvalue context then the tied 
object is going to have its STORE method called...

--
Piers

Thanks.

I'm new to the discussion of Perl6, so are there any discussions around 
providing operators such as wantlvalue and wantvoid
to perform similar queries to wantarray?


Ian
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: lvalue subroutines

2001-02-27 Thread Ian Brayshaw

Robin Houston wrote:
On Tue, Feb 27, 2001 at 11:38:40PM +1100, Ian Brayshaw wrote:
 
  I'm new to the discussion of Perl6, so are there any discussions
  around providing operators such as wantlvalue and wantvoid
  to perform similar queries to wantarray?

Yes. Damian has proposed
   http://dev.perl.org/rfc/21.html

which seems sufficiently generic

Cool.


PS:  You know that "wantvoid" is just !defined(wantarray()) ?

ah, yes, RTFM...


Ian
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




Re: lvalue subroutines

2001-02-27 Thread Ian Brayshaw

On Tue, Feb 27, 2001 at 10:50:26PM +1100, Ian Brayshaw wrote:
  Given the following lvalue subroutine
 
  sub mysub : lvalue {
  $value;
  }
 
  is there any way for mysub() to be able to determine that it
  was called in an lvalue context?

Yeah there is, but you're not going to like it :-)

... omitted for sanity ...

Thank goodness for TMTOWTDI (i.e. if in doubt, rewrite the problem)

Cheers.
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.




RE: In defence of Perl

2001-02-13 Thread Ian Brayshaw

  Is there any benchmarking available of salary bands for
  differing skills,
  i.e. are perl guys much more than ASP guys who can also do
  the other bits ?
  Any other arguments I should be making !

Yeah, Perl people cost more than ASP people and they're frickin
impossible to find, and vary wildly in quality. This is IMHO Perl's
single greatest barrier to acceptance, and in business terms it's an
entirely valid point.


From my experience, in the current contracting market ASP coders are more 
expensive then Perl developers (by around 10-15%). The variation in quality 
of output would hold for both sides. IMHO Perl developers tend to know more 
about good coding practices than VB/ASP coders (whether they practice them 
is another matter) since VB isn't exactly a difficult language and pretty 
much anyone can pick it up quickly. This could be a factor with long term 
development/maintenance if you have staff turnover.


Ian
_
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.