From mboxrd@z Thu Jan 1 00:00:00 1970 From: Timothy Sample Subject: Re: 05/05: gnu: Add ghc-hspec-discover. Date: Sun, 09 Feb 2020 11:25:57 -0500 Message-ID: <87sgjjpw16.fsf@ngyro.com> References: <20200208144029.3072.2814@vcs0.savannah.gnu.org> <20200208144032.CCDFC21049@vcs0.savannah.gnu.org> <87wo8wpbfj.fsf@ngyro.com> <87y2tcxh3y.fsf@nicolasgoaziou.fr> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Return-path: Received: from eggs.gnu.org ([2001:470:142:3::10]:56791) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1j0pPG-0003fr-L9 for guix-devel@gnu.org; Sun, 09 Feb 2020 11:26:05 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1j0pPE-0006Fp-OW for guix-devel@gnu.org; Sun, 09 Feb 2020 11:26:02 -0500 Received: from wout3-smtp.messagingengine.com ([64.147.123.19]:50583) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1j0pPE-0006FT-D4 for guix-devel@gnu.org; Sun, 09 Feb 2020 11:26:00 -0500 In-Reply-To: <87y2tcxh3y.fsf@nicolasgoaziou.fr> (Nicolas Goaziou's message of "Sun, 09 Feb 2020 10:08:49 +0100") List-Id: "Development of GNU Guix and the GNU System distribution." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-devel-bounces+gcggd-guix-devel=m.gmane-mx.org@gnu.org Sender: "Guix-devel" To: Nicolas Goaziou Cc: guix-devel@gnu.org Hi, Nicolas Goaziou writes: > > Timothy Sample writes: > >> We already have =E2=80=9Chspec-discover=E2=80=9D without the =E2=80=9Cgh= c=E2=80=9D prefix. The two >> packages look identical to me =E2=80=93 is this an unintentional copy? > > Oh! I hadn't noticed! > > Well, is there any strong reason to use "hspec-discover" over > "ghc-hspec-discover", besides that it already exists? The loose convention is that libraries have the prefix, and programs don=E2=80=99t. The =E2=80=9Chspec-discover=E2=80=9D package was added befo= re my time, but I guess whoever added it was following that convention. > For the arguments in favor of using ghc-hspec-discover: > - every other "hspec" package uses the "ghc-" prefix, > - the hackage importer names it, > - the hackage importer will add "ghc-hspec-discover" as an input anyway. That is definitely a problem. From my experience, =E2=80=9Chspec-discover= =E2=80=9D is often missed by the importer because it is put in =E2=80=9Cbuild-tool-depen= ds=E2=80=9D instead of =E2=80=9Cbuild-depends=E2=80=9D in the Cabal file. Ideally, the= importer would know to add it from =E2=80=9Cbuild-tools-discover=E2=80=9D and it wou= ld know what it should be named. > So what about deprecating "hspec-discover" in favor of > "ghc-hspec-discover"? I did a quick scan of the =E2=80=9Ctext-conversions=E2=80=9D source code, a= nd it doesn=E2=80=99t seem to use =E2=80=9Chspec-discover=E2=80=9D as a library. It looks like i= t=E2=80=99s just a mistake in the Cabal file. Between the convention and the fact that =E2=80=9Chspec-discover=E2=80=9D i= s almost always used as a program instead of a library, I would say that it=E2=80=99s okay the way it is. Beyond that, it would be quite a bit of churn to change it with very little benefit. > Thanks for the heads up! No problem. Thanks for committing packages! -- Tim