Scraping: Web Based vs Desktop Based

It really depends on the industry and what the actual scraper does. Web-based app costs WAY more and if the data scraped can't be shared (example: Scrapebox) it will be nightmare running it as a web-based scraper.

p/s: Any idea what happened to my previous thread that you commented on?

I'm not using server nor desktop scrapping.

I'm always writing scrappers for *nix based systems, so that they are written in python/perl/php
They runs on my home router. which is the last step before the internet.
The router has a script which reset it every 1 day, and the IP changes.

The development is done under windows machine and then transfered to the router.
I have an HTTP server with small system which allows me to manage the cron jobs and stop/start scrpping jobs from anywhere.
The router has a DNS name which it updates.

I am a veggie when it comes to tasks, especially scraping tasks.

I use flower for celery.

I agree with boberbrian.

It depends on your ultimate intent. Do you plan on releasing a scraper to clients? Are you just throwing together a scraper for some websites or are we talking about a generic, heavy-duty scraper?

I like to build my scrapers in *nix environments with python, it's quick and it works like a charm thanks to awesome python libraries.

If you think about developing a scraper which you'd like to orchestrate to clients, go for web-apps or a client-server model.