Re: [Lumiera-work] INFRASTRUCTURE build and test system on the server

Michael [Plouj] Ploujnikov wrote:
> Christian and I already started playing around with the builddrone
> implementations
> (http://git.lumiera.org/gitweb?p=builddrone/ct;a=summary and
> http://git.lumiera.org/gitweb?p=builddrone/plouj;a=summary).
> 
> I was trying to think how to implement a usable script for this
> system, but I ran into a few questions that I can't answer myself:
> 1. How do we organize or track relationships between functions in the
> soup directories which refer one to another. For example, both GitPull
> and Doxygen might use the AddToReport function. Should the AddToReport
> definition be duplicated in the files for both of the other files? How
> will Bash handle sourcing the same function twice? What happens if the
> two versions of AddToReport start to differ? Should we put AddToReport
> in a separate file which is sourced by the files where GitPull and
> Doxygen functions are defined?

Generally this soup shall only define functions and rather not execute 
anything else (it would be possible but can't rely on other functions 
then). Further, this all is very sketchy and has some rough edges which 
need to thought about. Thats more a design and definition issue.

For example: each scriptlet shall log its output in its own log 
(,git.log, ,doxygen.log, ...) actually we might even define that all 
this logs have some well defined format. My suggestion would be asciidoc 
which in the simplest case just places all in a listing block:
echo -e ".Git Pull Log\n-----" >,git.log
echo -e ".Git Pull Errors\n-----" >,git.err
git pull 1>>,git.log 2>>,git.err
echo "-----" >,git.log