I too would find variable manipulation to be handy.  I'm in the
process if finding a work-around now.

Randy


On Jan 20, 3:27 am, Ruben Willems <[email protected]> wrote:
> Hi
>
> now I understand it better, but on how to implement it,
> I have to think it over ;-)
> do not have a solution in mind right away.
>
> with kind regards
> Ruben Willems
>
> On Tue, Jan 20, 2009 at 9:23 AM, CinnamonDonkey <
>
>
>
> [email protected]> wrote:
>
> > Hi Ruben,
>
> > You are right... there are a lot of Cruise Control specific variables
> > already exposed and they are very handy too :-). I'm using at least
> > half of them in my scripts ;-)
>
> > I was just thinking that form a USER data point of view, it would be
> > nice to have a mechanisim by which the user could hold the odd bit of
> > data in one place along with all the other state information. The
> > alternative is to hand craft more complex scripts and to have them
> > create yet more files and access/update these files on every run.
>
> > One potential issue with this method is that the data being held could
> > become out of sync. Lets say there are 3 build tasks (I actually have
> > 5 in a full build run) task 1 runs and updates the file, task 2 runs
> > and fails (or worse, locks and gets terminated), task three never gets
> > to run. The users variable file is now out of sync holding data that
> > could corrupt the next build (depending on what the build process is
> > and what the data represents).
>
> > Cruise Control can be thought of as being Atomic, in that, it knows
> > when the build has started and it knows for sure when it has finished.
> > It could create a enviroment variable on behalf of the usere at the
> > start of the build process representing the project user state data,
> > held in the .state file and then at the end of the process write the
> > value left in the enviroment variable back to the state file so that
> > it may persist between builds.
>
> > Addition options could be available to specify clean up of the
> > variable on a failed build say, or a default value should the variable
> > not yet exist (perhaps because the .state file does not yet exist).
>
> > It's just a thought,
> > Shaun
>
> > On 19 Jan, 15:34, Ruben Willems <[email protected]> wrote:
> > > Hi
>
> > > there are already a lot of data exposed to the calling executables
> > > these are the integration properties.
> >http://confluence.public.thoughtworks.org/display/CCNET/Integration+P...
>
> > > for example the exec task gets these via environment variables,
> > > Nant and msbuild gets these as propertieshttp://
> > confluence.public.thoughtworks.org/display/CCNET/NAnt+Taskhttp://confluence.public.thoughtworks.org/display/CCNET/MsBuild+Task
>
> > > with kind regards
> > > Ruben Willems
>
> > > On Mon, Jan 19, 2009 at 4:25 PM, CinnamonDonkey <
>
> > > [email protected]> wrote:
>
> > > > Hi All,
>
> > > > I was wondering, is it possible to include user defined state
> > > > variables in the .state file?
>
> > > > If not, this would be a handy feature. The ability to define a State
> > > > Variable in the Project block which is loaded as an enviroment
> > > > variable and therefore accessable/modifyable by external applications.
> > > > The enviroment variable could be initialised from the .state file at
> > > > the start of a build and then written back to the .state file at the
> > > > end of a build.
>
> > > > What do people think?
>
> > > > Cheers,
> > > > Shaun- Hide quoted text -
>
> - Show quoted text -

Reply via email to