Brad Chapman wrote: > I wrote some documentation (now out of date, of course) about how this > works, which can be found at: > > http://www.bioinformatics.org/piper/documentation/piper_plugin.pdf You could have spared yourself a lengthy email and just gave me this link Brad :) This is what I understand of Alexandre's and Brad's e-mail about Narval's and Piper's pattern to describe 'applications': both methods are more or less the same in the aspect of both focussing on the application as a unix user in a shell would do. The name, input and output are well handled. I will wait for this to become more detailed once we can do some testing, and as Jeff said: we should not touch the stdin\out. But at the same time I see some critical issues missing, which are noticed by Brad and documented by some ?'s. The first is 'environment description'. By this I mean things like path (-relocation), dependencies and local issues (like hardware dependent stuff). The 2nd is about errors. I think Piper needs some basic form of exception handling, this can be just a simple retry of node execution. Maybe the PL already has something like this build in. (JeanMarc?). The 3th is event handling. Many applications only make sense to execute when there an special situation. Anybody has thoughts about these three points I feel missing? bye, jarl