I think requiring people to have ruby installed would be a fairly large hurdle to the adoption of rake. Sean Carpenter
On Tue, Oct 28, 2008 at 9:39 AM, Gabriel Schenker <[EMAIL PROTECTED]>wrote: > > I just had a discussion with Fabio about the fact that the current > build process for NHibernate is fragile and error prone. We have far > too many situations where all of the sudden the build breaks. Some > reasons include > > - the OS (Windows XP, VISTA, Server 2003 or 2008 32bit versus 64bit) > - Visual Studio (VS 2005 versus 2008) > - .NET framework (.NET 2.0 versus 3.5) > - nant (do I have the right version on my system and is the > environment path variable configured correctly) > - tests (do I have an SQL Server and do I have created the [test] and > [test2] databases > - human errors: some times a committer just checks in some thing that > only "works on his machine" > - etc. > > all these factors accumulate to a situation which is far from optimal! > It's a while ago but I have written an article about how to configure/ > prepare the system for NHibernate. > > > http://blogs.hibernatingrhinos.com/nhibernate/archive/2008/03/31/prepare-your-system-for-nhibernate.aspx > > There I described how to download the latest source from trunk and > build it as well as for .NET 2.0 as for .NET 3.5. > But today I tried to follow this step-by-step guide and I miserably > failed... :-( (not for the first time!) > > I think that we can agree that we have now reached a broad adoption of > NHibernate. The more people are using NHibernate the better the build > has to be (did I hear "friction-less"?). And yes I am aware that most > of the people just take the available GA (currently 2.0) but there is > still quite a big number of people living on the trunk (myself > included). > > As Fabio told me: "..nobody is willing to (really) maintain the nant > based build..." > > My suggestion is now to completely re-work the build of NHibernate and > implement it e.g. with a "better" build tool that xml-based nant or > msbuild files. My suggestion would be "rake" (other suggestions are > welcome). > In a first phase we can run this new build in parallel to the current > nant based build until every body can feel the difference > (by the way: the guys maintaining fluent-nhibernate did already base > their build on rake... ;-) ) > > Before I start implementing something I would love to hear some > feedback about this. > > a) do you agree that the current situation is a pita? > b) do you think completely re-working the build and use a different > tool (e.g. rake) is adequate? > c) did I forget some thing? > d) are you willing to help?
