gump-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Clayton Harbour" <>
Subject RE: [nant-dev] Re: NAnt
Date Mon, 06 Sep 2004 20:10:54 GMT
Hi Jarek,

Absolutely nothing :-).  

I was talking with Adam the other day and he suggested that a batch/ shell script that built
NAnt from source might be a little better than relying on the mini-NAnt.exe.  I believe ant
uses something similar to bootstrap itself so my understanding is that this fits in better
with the gump philosophy.

My thoughts are that the process is designed to take away the chance that something "different"
was done to generate the mini-NAnt binary.  I assume something like this just helps to better
capture the full build process...but I could just be talking bunk :-).

As far as the bin/NAnt and bin\NAnt command lines, I use them all the time so I would not
recommend changing that.



-----Original Message-----
From:	Jaroslaw Kowalski []
Sent:	Mon 9/6/2004 12:51 PM
To:	Clayton Harbour; Adam R. B. Jack;;
Subject:	Re: [nant-dev] Re: NAnt
Hi Clayton!

What's wrong with using the NAnt-provided bootstrap procedure? It uses a
portable mini-nant that is stored in CVS can run on any platform supported.

You just do either:



mono bin/NAnt.exe

and it will build the entire nant and put it into build/**/bin
If you want to build external dependencies, just compile them and put in the
appropriate "bin/lib" subdirectory before bootstrapping NAnt.

----- Original Message ----- 
From: "Clayton Harbour" <>
To: "Adam R. B. Jack" <>;
<>; <>
Sent: Monday, September 06, 2004 9:34 PM
Subject: RE: [nant-dev] Re: NAnt

Hi Adam,

I managed to get a first cut at the bootstrap together.  I have only
produced the Windows batch file and just want you to take a look and make
sure it is something you can use before I work on the ?nix script.

There is a batch file attached (extension = .bad so hopefully it won't be
stripped off) that will compile a minimal NAnt and then build itself from
that.  Just some notes on the setup:

    COMPILER_EXE: Full path to your compiler (these are defaulted to the
default install locations if not specified).

    LIB_REPOSITORY_DIR: Points to a directory with the contents of the bin
directory, namely (these will eventually be built if I understand





        NOTE: scvs.exe is not required to build NAnt

    NANT_CVS_DIR: Directory of the fresh NAnt checkout.

There are a couple of other parameters I am playing around with, namely
FRAMEWORK_TYPE and FRAMEWORK_VERSION.  I will probably add these when I can
get mono to compile.  (Almost there, I am getting an exception that makes me
think it is grabbing the System.Web.dll from the ms framework instead of the
mono I might need to add a full path to that.)

Anyway, please take a look at the scripts and let me know if there are
naming conventions or anything else that should be sync'd up.

Also will the bootstrap sit in the gump repository or should it sit in the
project (i.e. NAnt) repository?



-----Original Message-----
From: Adam R. B. Jack []
Sent: Mon 9/6/2004 11:11 AM
Subject: Re: [nant-dev] Re: NAnt
Stefan wrote:

> In Gump's case we probably can live with making sure that "our" version
> is used instead of the one in the GAC and DEVPATH seems to be a way to
> achieve that (I've only read the first paragraph of that article,
> though 8-).

Clayton Harbour and I chatted on this weekend, and
with Clayton's help we had a simple build of NAnt working (using an
installed NAnt on the %PATH%).

The Gump code ought attempt to generate an assemblies path, and as of a
moment ago attempts to pass it as DEVPATH. I've not entered sufficient
metadata to test this out, but hopefully we can add a log4net or similar to
see soon enough.

BTW: I made three quick tweaks to the Gump metadata. I'll document them once
we settle on them.

1) A project can have an attribute of language="csharp" (and this is needed
to generate lib path).

2) The workspace can have <dotnet framework="mono-1.0.1" /> to set the
framework (passed to NAnt, must be a recently nightly, using -t:).

3) <jar ( ought be able to
be replaced by <output or <assembly now.

Theoretically (once we add more metadata) one ought be able to be in the
Gump SVN root and do: -w metadata/gump-n.xml log4net  (with or without --debug)

to see things working.



This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
nant-developers mailing list

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message