From mboxrd@z Thu Jan 1 00:00:00 1970 From: Danny Milosavljevic Subject: bug#35542: Installer does not display full backtrace on error Date: Sat, 4 May 2019 19:06:17 +0200 Message-ID: <20190504190617.1b8a8efa@scratchpost.org> References: <20190503085927.ib5gudynudeknclm@pelzflorian.localdomain> <87ef5fetnv.fsf@gnu.org> <8736lvka3k.fsf@gmail.com> <20190503174633.otzjizlfjhr4bkca@pelzflorian.localdomain> <87y33msgmy.fsf@gmail.com> <20190504122403.noe5hirk6u3qtmcm@pelzflorian.localdomain> <20190504145409.00768bf6@scratchpost.org> <87pnoy42o7.fsf@gmail.com> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha256; boundary="Sig_/hUManB7/ql+7CL/4eZhUaVY"; protocol="application/pgp-signature" Return-path: Received: from eggs.gnu.org ([209.51.188.92]:54831) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1hMy7r-0007EI-KJ for bug-guix@gnu.org; Sat, 04 May 2019 13:07:04 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1hMy7q-0004SD-Oo for bug-guix@gnu.org; Sat, 04 May 2019 13:07:03 -0400 Received: from debbugs.gnu.org ([209.51.188.43]:38581) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1hMy7q-0004S1-LT for bug-guix@gnu.org; Sat, 04 May 2019 13:07:02 -0400 Received: from Debian-debbugs by debbugs.gnu.org with local (Exim 4.84_2) (envelope-from ) id 1hMy7q-0002ow-CQ for bug-guix@gnu.org; Sat, 04 May 2019 13:07:02 -0400 Sender: "Debbugs-submit" Resent-Message-ID: In-Reply-To: <87pnoy42o7.fsf@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: Mathieu Othacehe Cc: 35542@debbugs.gnu.org --Sig_/hUManB7/ql+7CL/4eZhUaVY Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable Hi Mathieu, On Sat, 04 May 2019 18:14:48 +0200 Mathieu Othacehe wrote: > Thanks for testing Florian. Based on Danny suggestion we could set > COLUMNS to 200 to have a more verbose backtrace (like guix itself). Good idea in principle. > Using higher values, the backtrace becomes really long and harder to > read. WDYT? When we have a designed error message, we present one--and, there, brevity is good. But backtraces are for developers only anyway, and are not designed error messages. So for backtraces I'm against throwing *anything* of the backtrace message away because we don't know what part of it is important (if we did, we'd print a designed error message instead--I'm all for doing that). I think this is something the default Guile setting gets wrong and I don't understand how it can be a good idea to suppress half of a message intended to simplify debugging for a developer "because it's too long". The length is nothing compared to the amount of text a developer has to type in order to reproduce the missing parts later on, if it's even possible. I'm even actually using core files (including attaching to them using a debugger in order to examine variable values), so I might be an outlier. So I'd be for either presenting the entire backtrace or suppressing the entire backtrace, not having teaser backtraces :) --Sig_/hUManB7/ql+7CL/4eZhUaVY Content-Type: application/pgp-signature Content-Description: OpenPGP digital signature -----BEGIN PGP SIGNATURE----- iQEzBAEBCAAdFiEEds7GsXJ0tGXALbPZ5xo1VCwwuqUFAlzNxokACgkQ5xo1VCww uqWrtggAjM1wnHPwxkPEpC5vjdrIj45hnnsyPK5LieaMAkdbbFceFtBs65Hbsnvu FFdoqcsVJMnpAjx4iBSMKnh+qdofE3A9MiRjcgNfRe1cxfENvtefcrPb/6iq5goT UN83bhqeQ/DMzmpKDY/Io/NBA65nFRPnmtxo7LYVXiw8RUdFjAX26pYwuAqQyE+4 ygeyhKA3mJZQO9Ekuusi0sxkPjwbGqVaovPiMU4248LtKYwd8xY5QgldLaPhj5AE EG2lWCMhBG5IOuP/DfZoK6Z2gwIWzxTo8wYDIZFo0+cRvyJOKUtVerNbgsmwMjhw HMO2DdbZMq2sAUhF/xltMS2n1Ppn2w== =m2Xv -----END PGP SIGNATURE----- --Sig_/hUManB7/ql+7CL/4eZhUaVY--