Multiple idatabase registration windsor how to#
Multiple idatabase registration windsor software#
Multiple idatabase registration windsor code#

When it comes to multi threading, logging is not an easy task.There a certainly a lot of people outthere who are more focussed on that topic and have more experience.Writing log libraries is not our core competence.Sure we could do that, but I guess there are a lot of reasons against it: Should we create a log library by ourselves to have it completely under control? Verbose information shall only be logged when necessary. Five levels of logging are pretty usual: Debug (aka verbose), Information, Warning, Error, Fatal. The log shall not make a programm less performant when it runs on a customer’s machine. We need to put out information in different levels. The most important information from my perspective if the complete execution path of a program.

Multiple idatabase registration windsor how to#

For more of the reasons have a look here at How to do sensible logging in c# We wouldn’t have any possibility to find out what is going wrong on customer site without it. Most of the questions had been answered pretty fast:

Multiple idatabase registration windsor code#

Or let’s say, how the code taught me what I did wrong so I needed to do things differently to survive 😉 You may have a look onto Evolution of methodic where I described how I developed a certain set of rules. The coding principles are mostly SOLID, plus some sugar.We’ve chosen Windsor Castle as Dependency Injection framework.The answers to the questions above in just a second We’ve chosen log4net as the logging library.

Multiple idatabase registration windsor software#

We had to decide which technologies to use, what the basic infrastructure for this piece of software is, what coding principles come into the play. Pretty considerable requirements, let me introduce the most problematic one: It shall automate the “real” Excel application running as a service. Is shall be capable of starting job on a scheduled base.It shall have a web-based API to remote control it.If parts of a certain processing fail, not the complete processing shall be restarted, but the pieces which failed Being able to scale out with multiple machines.Being able to scale up with multi-threading.Some requirements had been changed, so it should be recreated with a lot of constraints that the full client couldn’t offer: About three years ago, pmOne AG, the company I do work for, decided to completely rebuild a certain piece of software. Let me come up with a little background information. How to put the hurdles of logging that deep that nobody will complain about doing it?.Does AOP pollutes the code, when I add interfaces/ virtual methods for logging reasons?.Does a log message pollutes the actual code?.Should we create a log library by ourselves to have it completely under control?.😉 That was a longer discussion, and a lot of questions had been raised: Quite a while ago I had a good talk with an colleague about “to log or not to log – that is the question”.