Date: Sat, 16 Jun 2007 16:28:19 -0400 From: Kris Kennaway <kris@obsecurity.org> To: Stephen Hurd <shurd@sasktel.net> Cc: ports@FreeBSD.org, Sam Lawrance <boris@brooknet.com.au> Subject: Re: Clarification on fetch/extract targets Message-ID: <20070616202819.GA29331@rot13.obsecurity.org> In-Reply-To: <46742CF6.3050901@sasktel.net> References: <4673B0DB.3040100@sasktel.net> <A067E197-1540-43E6-B396-7F08179C0CA2@brooknet.com.au> <46742CF6.3050901@sasktel.net>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --] On Sat, Jun 16, 2007 at 11:33:26AM -0700, Stephen Hurd wrote: > Sam Lawrance wrote: > >>So, I created a new port (ports/113608) for Code::Blocks which has > >>not had a source code release for almost two years despite constant > >>development. However, the currently correct way to get the source is > >>via subversion. The port currently does this, but I'm unsure of > >>exactly what is required by the fetch and extract targets, so I'm not > >>sure what the "right" thing to do here is. I can think of a few > >>possibilities: > >> > >>1) The fetch target is intended to do all the steps which require a > >>network connection and extract is intended to places the files into > >>WRKSRC > >>- In this case, the fetch target would use svn co to WRKSRC and the > >>extract target would do nothing. This is what the port currently does. > >> > >>2) The fetch target is intended to place source archives into DISTDIR > >>and extract is intended to place them into WRKSRC > >>- In this case, the extract target would use svn co to WRKSRC and the > >>fetch target would do nothing. > >> > >>3) The fetch target is intended to place a source archive in DISTDIR > >>and the extract target is intended to extract that archive into WRKSRC > >>- Here I would need to do the svn checkout in the fetch target, > >>archive it, then delete it. > >> > >>4) The fetch target retrieves sources in the distribution format and > >>extract places them into WRKSRC > >>- In this case, you would svn co to a DIST_SUBDIR in fetch, then > >>extract would copy that to WRKSRC > >> > >>5) fetch *MUST* fetch a source archive into DISTDIR. None of this > >>svn/CVS checkout stuff in the ports tree. > >>- Hrmph. So rather than bumping PORTREVISION to the desired > >>subversion revision and having it Just Work, one needs to make a > >>source archive and host it somewhere himself and be responsible for > >>updating it himself too. > > > >In the past, it has been easiest if the maintainer (you) creates an > >archive of the source and either puts it up for download, or asks for > >it to be mirrored as a local distfile (the committer who deals with > >the PR will do this for you). > > > >To make it easier for yourself, you might create a "roll-tarball" > >target in the port, which you (or future maintainers) can use to > >automatically fetch the source from subversion and create a new source > >archive. > > Actually, I found it quite easy to have the port pull the sources from > svn. Who are we concerned about making it easier for and why (and how > is it any easier?) Everyone behind a firewall that only allows fetching via HTTP/FTP, for one. Also everyone without live network access, and those with pay-per-download who have a free local distfile mirror, etc. Tarballs are overwhelmingly preferred. Kris [-- Attachment #2 --] -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.3 (FreeBSD) iD8DBQFGdEfjWry0BWjoQKURAs67AJ47TV+qpsRxhF5OsAeKzqzUhPnhXgCg8pbi fKdknO6nTtan0ak7WGrC2f0= =pVEA -----END PGP SIGNATURE-----
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070616202819.GA29331>
