From mboxrd@z Thu Jan 1 00:00:00 1970 From: Danny Milosavljevic Subject: bug#37286: Make a faster method to list supported boards. Date: Mon, 2 Sep 2019 22:49:04 +0200 Message-ID: <20190902224904.57e41fa4@scratchpost.org> References: <5f87e4c297b8e4fca0f9ceb0729f2176dec947bb.camel@gmail.com> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha256; boundary="Sig_/116EnjlWSjHOkB9cCzS/RnQ"; protocol="application/pgp-signature" Return-path: Received: from eggs.gnu.org ([2001:470:142:3::10]:60425) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1i4tH2-0003Z0-9o for bug-guix@gnu.org; Mon, 02 Sep 2019 16:50:06 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1i4tH1-0007Ot-4m for bug-guix@gnu.org; Mon, 02 Sep 2019 16:50:04 -0400 Received: from debbugs.gnu.org ([209.51.188.43]:50201) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1i4tH1-0007Ok-1t for bug-guix@gnu.org; Mon, 02 Sep 2019 16:50:03 -0400 Received: from Debian-debbugs by debbugs.gnu.org with local (Exim 4.84_2) (envelope-from ) id 1i4tH0-0004qe-5O for bug-guix@gnu.org; Mon, 02 Sep 2019 16:50:02 -0400 Sender: "Debbugs-submit" Resent-Message-ID: In-Reply-To: <5f87e4c297b8e4fca0f9ceb0729f2176dec947bb.camel@gmail.com> List-Id: Bug reports for GNU Guix List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: bug-guix-bounces+gcggb-bug-guix=m.gmane.org@gnu.org Sender: "bug-Guix" To: Jesse Gibbons Cc: 37286@debbugs.gnu.org --Sig_/116EnjlWSjHOkB9cCzS/RnQ Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable Hi, On Mon, 02 Sep 2019 12:52:49 -0600 Jesse Gibbons wrote: > guix tries to build an entire system. It doesn't look like it checks if > "dne-board" is a valid board until it is building the image. It takes > hours to cross-compile the kernel, and even longer if the kernel needs > to be deblobbed. Yes, that's true. For better or for worse, there are a LOT of different ARM boards. Back when I implemented that part I thought that when you want to find out whether your board is supported, there's a good chance that you want to install Guix anyway--so you need the system image anyway. If it turns out not to be supported you just uselessly built a lot of stuff--but the substitute cache should have cached all that stuff anyway. But for some reason the ARM build farm substitute cache has a very bad hit rate (I remember waiting MONTHS to finally get a "flash-image" substitute that I didn't build myself). If we decided to do so, we could limit ourselves to just a few that we specially support--but that would make Guix System really a non-universal operating system. (with the current state of Guix ARM implementation it is anyway) It would be easy to get the list of supported u-boot targets from the derivation. It might be that Guix Data Service would help with that (see "More progress with the Guix Data Service" by Christopher Baines). Or we can maintain a list in the package definition ourselves. What do you think? --Sig_/116EnjlWSjHOkB9cCzS/RnQ Content-Type: application/pgp-signature Content-Description: OpenPGP digital signature -----BEGIN PGP SIGNATURE----- iQEzBAEBCAAdFiEEds7GsXJ0tGXALbPZ5xo1VCwwuqUFAl1tgEAACgkQ5xo1VCww uqXbSgf/WkxKl80pukLaejveXy6jRxEIq33uSWCMII7FPgfmyiiMJofxeQSdXufo IC1/ZlLtawlKdsGQFbTj3w0ytXmEjkJe8i9AmxJ9ZcElikSA6tSV1Jn3iQJ+ZWDC N+YMaa5D8PfARo3FSZmyQn9Nevyluw7e5AVvOpXZLmP3oLthUA0rC4WtxtffQ4kg 4JTtq3ppDRHagthv4QgPokJQbkntkR1vLTyCSErLKjmO/6r2bVFvuamdA8PF7c+x eNErz6v0MaqL0kaSUU2ZsscLgmcfiS1u8GQy4BiRaKUyDa/MZnYFTKf+ryFRqj5o BQBD3ca7GBd1Mgam3+vw8FOg4kyAFg== =aF01 -----END PGP SIGNATURE----- --Sig_/116EnjlWSjHOkB9cCzS/RnQ--