Comments on: Explicit or implicit?

By: Tau

Tau — Mon, 19 May 2008 21:47:49 +0000

I’m embracing Python myself because of Google’s AppEngine… and I agree – among other languages, it’s one of the most mature and robust. The only part I dislike with Python – its whitespace paradigm…

By: Stodge

Stodge — Wed, 14 May 2008 16:21:23 +0000

Haha! I loved the celebrity example in Joel’s article, though I’d gladly let most of them travel via IP. I won’t complain if 95% of all Hollywood actors don’t arrive at their destination.

Interesting article – I love abstraction, or packing in the sense of presenting a clean API to a developer. But I think I’ll have to spend more time thinking about “leaky abstractions” as I’m currently prototyping an in house API.

By: Steve

Steve — Wed, 14 May 2008 09:04:19 +0000

@Randal: interesting, I haven’t used Smalltalk in many, many years

@Mark: yeah, I also feel more comfortable with defining model state in my code rather than it being implied from the database. I think you should only base code on data for legacy dbs, for new systems it feels more natural the other way around.

By: Mark Ramm

Mark Ramm — Wed, 14 May 2008 00:48:33 +0000

I agree. I did some rails work before joining the TurboGears project, and one of the the first things I noticed was that TG had explicit imports so I knew where to look to find out how things were being done, while in Rails you often had to know where to look before you could even find the code you needed to read to understand the magic.

I liked lots of things about rails, but convention over configuration shouldn’t be pushed so far.

By: Randal L. Schwartz

Randal L. Schwartz — Tue, 13 May 2008 21:37:46 +0000

Have you seen Seaside (http://seaside.st)? I think it’s even more elegant, and faster to develop, and faster to execute than RoR. And it’s written in Portable Smalltalk. Check it out.

By: Steve

Steve — Tue, 13 May 2008 21:25:29 +0000

Thanks, it seems once again I’m doomed to reach conclusions that others have raised several years ago

By: Sören Meyer-Eppler

Sören Meyer-Eppler — Tue, 13 May 2008 20:25:10 +0000

Your points remind me of Joel Spolsky’s article about “Leaky Abstractions”:

http://www.joelonsoftware.com/articles/LeakyAbstractions.html

regards,

Sören

By: Steve

Steve — Tue, 13 May 2008 20:10:17 +0000

Not wooly at all – whatever works for you. We all use our experiences to judge what’s best in a given circumstance.

Personally I’ve used quite a lot of closed libs & frameworks over the years and although it doesn’t happen that often, occasionally I’ve had serious blocker issues that the user community either didn’t know how to resolve, or I had an issue that was too specialised, and the official support guys (most often MS, Oracle or IBM) sometimes took a while to figure out – and even then it might be a ‘wait until the next fixpack’ kind of issue, with hacky workarounds in between. Generally with the open source stuff I’ve encountered fewer (or in most cases no) absolute blocker issues. That’s simply because there are more people familiar with the code – not just looking, but changing – and thus more variety of experience (and thus more chance someone who both knows the code and understands the issue is contactable). Worst case, I can always put a patch in myself if I’m in a tight spot. It’s led me to the conclusion that I just feel more comfortable that I won’t get stuck on an edge case when using open source. It doesn’t mean I won’t use closed source libs anymore, I just prefer not to if good alternatives are available.

By: Paul Evans

Paul Evans — Tue, 13 May 2008 19:14:52 +0000

Opinion? I’ve only ever been to talks about Ruby on Rails… I like the idea of how it versions databases. I like the English style 5.minutes.ago.

I guess as I work with C# a lot of the time I am one of those “happy to trust and move on” guys. There are good communities around C# and .Net things – and even if it’s not open source there is a lot of information available about it, and you can *look* at a lot of the source via reflector, or even step in to some c# framework source if you have VS2008.

C++ for the other part of my time, and I guess when I use libraries I tend to trust their function / object contracts and get on with it – and only check up on them when the source is available if something comes up through profiling or errors. In fact I could say the same of C#.

I guess I’m fairly pragmatic, I tend to use whatever is available and seems the best fit for what I’m doing at the time. Though writing that seems rather wooly.