Comments on: Source Services 0.0.1, no more writing SPEC files ?

By: Livio

Livio — Wed, 05 Aug 2009 11:19:10 +0000

Would be awesome to stop writing SPECs, they’re boring and useless. It takes a lot of time to write a SPEC. Packaging for Linux takes much more time than creating an installer, it needs to be quicker and smarter 🙂 .

By: Mike Cronquist

Mike Cronquist — Tue, 04 Aug 2009 20:47:25 +0000

Thanks. Testing comment field.

By: Andrés G. Aragoneses

Andrés G. Aragoneses — Mon, 03 Aug 2009 14:01:21 +0000

BTW, please indicate in your spam protection system that the number has to be written with digits!! I’ve been trying to put a comment like five times (writing things like “five” or “f-ive”, when I should have written “5”).

By: Andrés G. Aragoneses

Andrés G. Aragoneses — Mon, 03 Aug 2009 14:00:03 +0000

It’s very pleasant to finally see that some efforts go against duplication of efforts. I’ve always seen spec files and tools like jhbuild a way of doing the same thing twice.

By: Adrian Schröter

Adrian Schröter — Mon, 03 Aug 2009 13:39:31 +0000

and it would break use cases like downloading a -latest.tar.gz link.

By: Adrian Schröter

Adrian Schröter — Mon, 03 Aug 2009 13:38:03 +0000

As written my current thinking is that validating, eg. via a md5 (or better sha1) would be an extra service. I do not see why it must be mandatory in the downloader service, since all of this files became anyway part of the source repository. So you can also see afterwards when it has changed.

By: Srinidhi

Srinidhi — Mon, 03 Aug 2009 13:36:39 +0000

Hey Adrian! Very cool! I’m willing to help out.

By: Richard Guenther

Richard Guenther — Mon, 03 Aug 2009 13:31:32 +0000

Please add a mandatory field for a MD5 sum that is required to match that of the downloaded sources
(which could then of course also be used to implement a build-service-side caching mechanism of the
downloads, but that’s secondary).