[ic] Re:[mv]*** Interchange 4.6 & PgSQL Difficulties ***

BF meinbuch@mein-buch.com
Sun, 3 Dec 2000 10:46:12 -0500 (EST)

Quoting Mike Heins: 
> (I haven't pontificated for a while, and this seems like a good
> opportunity. )
>> Quoting Dan (db@list.dnsalias.net):
>> Mike, if Enhydra/Zope were already mature when minivend started
>> (another alternate universe question), would you have gone with a app 
>> framework or would you still have started your own framework, tag 
>> language, and parser?
> Possibly, but still unlikely. At one point early on I looked at PHP,
> which sort of fits in that realm, but it became obvious that building
> modules for it was too difficult to do. It was easy with Perl. And PHP
> seemed to require SQL for anything resembling a large catalog, which
> would mean tying the program to one particular database. Also, I was
> a novice programmer at the time and could barely write Perl code that
> passed syntax checks. 8-)
Hi folks,

don't be fooled... 8-). The last of above MH's sentences is
the most sympathetic little lie I ever heard, but Mike tends to say that,
so he could tell real beginners:  "If I can do it, everyone can do it",
which is of course also not true, but hey, comes in handy sometimes ...8-) 

This little paragraph about him, (was avaiable on the web in spring 1996
and still is today) just doesn't make  his "novice programmer" statement
sound right. 

quoting from: http://www.interex.org/conference/IWorks95/sis.html

	 Mike Heins has been involved in the computer world since 1977, 
	 and in UNIX systems since 1983. He once wrote a complete
         database in shell script, some modules of which were later coded
         in C. This system, which tracked company manufacturing, was
         in production use for over 4 years. 
         Heins has written articles for Instrumentation and Control
         Systems, Workstation News, Sun Observer, and other trade
         journals. He has presented papers at Uniforum, Sun User's Group,
         Buscon, and Sun  Developer conferences. He is currently employed
         as .... 

Of course I don't know when programmer among themselves call each other
a programmer, but I would say the above sheds some light on how much of
a "novice programmer" he was in 1995.  8-) 


> On another note, I still have yet to find anything which handles flat
> file databases with anywhere near the speed of Interchange -- they all
> pull the file apart and put it back together every time. If anyone has
> any pointers to something which does this well and handles 50,000
> records or more *fast*, I would like to see them.
> Nothing sells like speed, IMHO. If your shoppers are waiting perceptibly
> at every click, you are losing business.

Another gentle, sympathetic understatement. I agree with the last
sentence, but he dropped a zero in the above number.
(nothing sells like speed ... and NOT needing an expensive Oracle db for
ten thousands of records, MV was THE choice for the poor man early on).

The novice programmer Mike Heins handled 500 000 record flat files in
in summer 1996 with the speed of light (I think he worked on it six days,
and that mostly because to import, split, index and sort such huge
flat files took on simple machines around 16 hours and more). 

I think Mike Heins must he is a good card player and seldom reveals what
he has his hands. 

> The other problem with any of web development environments is getting
> access at a low enough level. When you start to *really* look under
> the hood of some of those things, you will find an awful lot of gotchas

> unless you tie yourself to one particular database. Which is why IC is
> the only true database-independent shop out there (that I know of).
> In fact, I can't think of any application at all which hooks to as many
> different database types without tweaking the core code. If I am proud
> of anything in Interchange, I am proud of that. 8-)

Right on, but now I have a question. On Akopia's website you can find
what is estimated a major site's implementation would cost:

                  $ $$$ $$$.cc

hmm, I remember I was impressed that one had NOT to spend $ $$ $$$.cc
for an Oracle db back in 1996 and that sold MV to me. 

I would prefer you would develop a commercial product and sell it
(may be like sendmail), so that the poor "Mensch" can still do something
at a price level which is more human and _buy_ a "box with some nice
toy" in it, which doesn't require the $ $$$ $$$.cc. help of expert
gurus ?

I know, I always tend to say the wrong thing at the wrong place at the
wrong time, but someone has to do that...so why not me : 

Can we expect a commercial product (with a GREAT documentation about
the internals in printed format) one day, which is around $ $$$.cc to 
$ 1 $$$.cc and poor "Mensch" still can afford to buy ?