[ProgClub list] Memory management in C

Stuart Laughlin stuart at bistrotech.net
Mon Oct 10 11:23:35 AEDT 2011

On Sat, Oct 8, 2011 at 11:28 PM, John Elliot <jj5 at progclub.org> wrote:
> I started doing a web-application in ASP.NET years ago, back in the 1.0 days
> when it was even more terrible than it still is, and I couldn't figure out
> how to design my app at all. So I ended up just implementing IHttpHandler
> and integrating with IIS/ASP.NET like that.

Probably for the best (and I vaguely remember hearing about that at
the time, come to think of it).

>> Then in late 2008 I went back to take a second look at Rails and
>> company and discovered that Django kicks Rails's ass (imho of course
>> ;). So in early 2009 I really dug into python/django, landed a big
>> eCommerce project, and spent most of 2009 and 2010 immersed in
>> django/debian/postgres/vim/etc. That project wrapped late last year.
> That sounds like a win to me. You got the project done in time and on budget
> and all of that?

I wish I could say that was the case, but actually the project was
fairly disastrous. It had nothing to do with the technologies, though.
The project was run by a guy employed by the client. He's a great guy,
and we're friends, but he didn't run the project well. And that wasn't
all his fault either; the client was an extremely dysfunctional
organization full of fiefdoms and power struggles. But I'm proud of
the work I did there. The system was ahead of its time, and it's a
shame they weren't ready for it.

> I've seen a little work in a Rails app that is trying to replace a whole
> line of business suite at my current client. Their previous programmer
> wanted to re-write the whole thing in Rails and MySQL and he made a bit of a
> start on it. I just don't know how that's feasible given all the SQL Server
> integration they do. I'm just not sure that a switch to a MySQL backend
> would be compatible enough with the way they work with their data. (I.e.
> integrating with Office and other Windows software. Although maybe the OLE
> DB (or ODBC?) driver for MySQL would be good enough to get the required
> level of connectivity.)

Yeah, it's hard to bring good non-Microsoft technologies into a purely
Microsoft environment. I'm partial to PostgreSQL over against MySQL.
I've done a little experimenting trying to integrate postgres into a
MSFT Office environment. A client was using an Access system that was
performing so horribly that it was essentially unusable --
three-minute waits between screens and that type of thing. I proposed
exporting the data to a postgres database and linking the Access
forms/reports to the data over there. I did a small prototype to
determine whether it was a viable solution (using the postgres odbc
driver for Windows to create the linked tables), and I was pleased
with the results. It seemed to work very well. In the end, the client
purchased some COTS software instead, so my prototype never saw a
production implementation.

> What sort of a company are you working for? Sounds like a consulting
> software development company, yes?

Right, the company has three practices:
networking/infrastructure/security, IBM iSeries, and software
development. The software development practice was the "Microsoft
Practice" when I started six years ago, then it evolved into the
"Business Intelligence Practice" (though I never really got into BI),
and now we are in the process of becoming the "Software Development
Practice" under the direction of my colleague and myself.

>> You should care about IoC because it forces you to think about how
>> your classes interrelate.
> But saying something like that is like saying nothing at all. *Of course* I
> think about how my classes interrelate, and I don't need an IoC framework to
> "force" me to do that! :P

Fair enough, but I think IoC leads you to think differently about it,
or at least that's what it did with me. It gave me a new/better/more
precise way of defining "single responsibility."

> I hear you say all those things and am still left wondering what on earth
> IoC is. :P
> Well it's probably going to be a *very* long time before I get around to
> actually using IoC. I've got the "design" of most of the software I intend
> to be working on for the next few years all nailed down (and it doesn't
> involve IoC) and now I'm just waiting to find the time to code it all up.

I intend to put together a couple/few emails this week explaining my
take on it. Let's see if my good intentions translate into actual


More information about the list mailing list