But, surely if utility classes are located in the same namespace or include file as the object they interact with this circular dependency is avoided?
Naturally this relies on developers understanding what they are actually including in a system, but if things are namespaced and named correctly then this task becomes easier.
37 signals describe a good example of a system exception they had been ignoring for ages that actually related to a feature of memcached. The issue had been ignored because it wasn’t affecting customers but it threw a development spanner in the works st a crucial tome.
The other problem with ignore the white noise of exceptions is that you can miss the real problems that only show up when you go live.
I think I’ve lived my whole life to this dictate. Except recently where I think am getting complacent, new years resolution becons me thinks. Remeber, your only as good as you make yourself, if you don’t have the skill, learn it, and you will.
I like this video, lot of insights. Like the karma points and the fact that they use bit torrents for replication.
One draw back about sprints and backlogs is that you always think about the other tickets and therfore do more than is required without thinking. If debate had no idea what was next except that there was a ticket they would concentrate on that one feature on work on every aspect until it was released.
Of course this only works if you have a good prioritised feature backlog.
Of course it’s not just them that do this, I’m sure Google know more about you because of analytics than you would like.
It’s a pity the UK legislations on cookies use will mean that many UK companies will be too scared to pay this much attention to their users.
I suppose the question should be, what do you loose if your company doesn’t learn much about your users. Can you really give them the best service of you don’t know what they do outside of.your website?