Date: Wed, 1 Dec 2021 21:28:36 -0800 (PST) From: Don Lewis <truckman@FreeBSD.org> To: Cy Schubert <Cy.Schubert@cschubert.com> Cc: Alexey Dokuchaev <danfe@freebsd.org>, Kubilay Kocak <koobs@freebsd.org>, Cy Schubert <cy@freebsd.org>, ports-committers@freebsd.org, dev-commits-ports-all@freebsd.org, dev-commits-ports-main@freebsd.org Subject: Re: git: b8c4bfe660b3 - main - sysutils/reptyr: Address LLVM 13 build failure Message-ID: <tkrat.58df421d89b81b69@FreeBSD.org> References: <202112020054.1B20sZXk040844@gitrepo.freebsd.org> <e9314dc6-5db0-5c01-96e7-4797f367835c@FreeBSD.org> <YagxPMhGLs%2BrFO5A@FreeBSD.org> <202112020412.1B24CA8G016980@slippy.cwsent.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 1 Dec, Cy Schubert wrote: > In message <YagxPMhGLs+rFO5A@FreeBSD.org>, Alexey Dokuchaev writes: >> On Thu, Dec 02, 2021 at 12:05:46PM +1100, Kubilay Kocak wrote: >> > On 2/12/2021 11:54 am, Cy Schubert wrote: >> > > commit b8c4bfe660b373862165a58514f270a51e77e147 >> > > >> > > sysutils/reptyr: Address LLVM 13 build failure >> > > ... >> > > 2 errors generated. >> > >> > Ports usually shouldn't use -Werror and people are strongly encouraged >> > to add -Wno-error >> >> On the other hand, every project really should enable -Werror so the >> compiler catches as much as it can for you. Having it enabled during >> development is a must, having it enabled in releases is arguable as >> it might bring extra burden to maintainer, but it does help to catch >> bugs in the environment original developer has no access to, so no, >> we don't strongly encourage -Wno-error when specific warnings can be >> scoped as it helps to catch other warnings in the future. Now they >> will likely went unnoticed, ergo unreported and unfixed. > > Then it behooves port maintainers to upstream patches whenever possible. > > This has been fixed using a proper patch and a pull request has been > submitted to our upstream. I should have done this from the get-go. This causes me some discomfort. I'm definitely onboard with working with upstream. My concern with local patches is that they can potentially introduce bugs, and now you are running code that has diverged from the presumably battle-tested upstream code. In one particular case, I'm working with some really crufty old C code that's been abandoned for quite some time. Compiling it throws tons of errors about things that were mentioned favorably in the original K&R C book. The code badly needs cleanup, much of which look like it is fairly mechanical, but it would be too easy to introduce bugs and there is no test harness that I'm aware of to do a thorough test of the updated code. Another case is a large C++ project where many of the errors are due to an API problem in a base class that is used by many other classes, some of which override the problematic method. Fixing the API issue in the base class ripples out to *many* other places and doing an adequate job of testing looks like a lot of work.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?tkrat.58df421d89b81b69>