Re: How is this something Borland isn't doing?
From: Liz (liz_wants_no_spam_at_xcalibur.nospam.co.uk)
Date: 12/18/04
- Next message: Uffe Kousgaard: "Re: Delphi 2005 demos"
- Previous message: Rudy Velthuis [TeamB]: "Re: God I Love Delphi7"
- In reply to: John Kaster (Borland): "Re: How is this something Borland isn't doing?"
- Next in thread: John Kaster (Borland): "Re: How is this something Borland isn't doing?"
- Reply: John Kaster (Borland): "Re: How is this something Borland isn't doing?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 18 Dec 2004 03:01:34 -0800
John Kaster (Borland) wrote:
> Try being a Borland employee and experiencing the same thing ;)
Ouch that has to hurt.
OK this next bit is long but I need to give you some huge background on
it..
A very very long time ago, when dos was still the main call of the
day.. windows and MS Access had just kind emerged..
I started a new job, they took me on coz Im an all rounder, and the guy
they had was.. um.. "unique" ..
Anyway, the process went.
Extract data from IBM mainframe
Extract data from PC Focus
Mix the 2 together using MS Access
Pump into convertion tool - which has no useful error codes
repeat Mix/Pump until it actually goes into the new format as you fix
things like the CSV data cant contain empty fields :(
Stick on a floppy, give to Client services to give to their clients.
This was done on a per week/month depending on the data type, and per
client basis..
When I started he was looking after like 4 clients, and it took him..
Try not to cry.. 4 weeks to do 1 dataset..
I found this somewhat, err, surprising.
However, less surprising when you find out, for each data/client type
he had, he didnt reuse last times program with maybe some adjustments,
he rewrote it from scratch all data extractions. The Access combine
could take 4 to 5 hours to run, and the pump could take upto (hahaha
this is so funny) 4 days.. sometimes more!
Anyway..
About a month after I joined he was off.. I got to actually do it "My"
way.. I did one in a couple of days.. I felt a lot better but I had
been able to put some thought process into it too.. And secretly worked
on a couple of simple things to make it even sleaker..
Next time the miserable man was off as I had talked to him about
changing it and he had refused claiming "I dont have time to test it or
even try it" despite the fact I had already run comparative tests as I
found I was generally able to keep up my couple of days level to his
3-4 weeks, so, I got to do most of the work (eg, Id do like 18 out of
20 clients) and my 18 would be done in just over a week, he would still
be doing his..
I processed 22 clients in less than 1 day.
How?
Well, the mainframe extraction rather than run 10 jobs with each
product in per client, I wrote 1 that did it all, and you just entered
the appropriate parameters. so, I only had 1 big file rather than 10
midget files to take from there.
The PC focus data extraction I used the last times one, as well, I had
kept them all :) Why rework the wheel.
I mixed the 2 files (not the 11+ files) in access.. as I hadnt won the
previous argument about foxpro vs access (more later)
I had written a small app in C to go through my data, I had changed
from using "," as my seperator, as people were found to be entering
addresses like "1, church street" and this upset our datapump, so, my C
program went through, if it found two seperators together it put in a
space, which is what our data needed, and if it found extra ones, it
told me.. It could run through our 50mb data in less than a minute, and
save the 20+ runs of the data pump to find out it failed.. The datapump
would then take about 4 hours or so.
I commendeered every spare machine in the building .. oh yes. Rar!
I tested each one as it finished, as I had put a novel send msg on the
end of each datapump, and it sent to me so I could then go to the
directory, get the dataset, check it, and hand it over.
My boss got back and was livid.. as well, people hadnt liked him much
anyway and even from when I started people thought i was his boss,
despite being like 18 or so at the time. But he was more livid because
it really did take him 3 to 4 weeks per client, and he couldnt handle
doing it any different. he quit. I didnt get his job as I had been
told I would, so I quit too ..
Anyway, onto the more relevant point, now you have an insight of my
work there, and the guy I worked with, I had suggested foxpro not
access.. Why? Well it was easier in foxpro to write a generic
extraction, one that would extract columns 1,3,4,5,6 etc.. rather than
having to use the exact names of the fields, which meant that each time
the months/weeks/quarters changed, the program had to, plus, it was
faster :)
My boss, showed me a magazine, in which it claimed many many false
things about foxpro for windows which was out then, it claimed you
couldnt change any colours, or fonts on the scren, it claimed you
couldnt use multiple files(!!!)... and despite being MS Foxpro by now,
my boss claimed "its not a microsoft product" .. so I lost there..
Funny, v2 of access of course stole a bunch of the technology Foxpro
had because they needed it to speed it up! ( When I say stole, they
bought the foxpro product specifically to use the technology in access
rather than any non legit method!)
- Next message: Uffe Kousgaard: "Re: Delphi 2005 demos"
- Previous message: Rudy Velthuis [TeamB]: "Re: God I Love Delphi7"
- In reply to: John Kaster (Borland): "Re: How is this something Borland isn't doing?"
- Next in thread: John Kaster (Borland): "Re: How is this something Borland isn't doing?"
- Reply: John Kaster (Borland): "Re: How is this something Borland isn't doing?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
