Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 30 Oct 2019 09:15:23 -0700
From:      Yuri <yuri@rawbw.com>
To:        Jason Bacon <bacon4000@gmail.com>, Mark Linimon <linimon@lonesome.com>, Gleb Popov <arrowd@freebsd.org>
Cc:        Mathieu Arnold <mat@freebsd.org>, Tobias Kortkamp <tobik@freebsd.org>, Raphael Kubo da Costa <rakuco@freebsd.org>, ports-committers@freebsd.org, svn-ports-all@freebsd.org, svn-ports-head@freebsd.org
Subject:   Re: svn commit: r515803 - head/devel/bazel
Message-ID:  <47ff19e3-b701-414f-c935-1c964c115f23@rawbw.com>
In-Reply-To: <ecbed237-1f89-f39b-1b7e-43656ad5beda@gmail.com>
References:  <201910271742.x9RHgvQP017264@repo.freebsd.org> <09d56af7-96f5-dcfd-d70c-d20d54e97ea2@rawbw.com> <20191027222033.GA81623@urd.tobik.me> <778754a2-2aa5-831a-7441-a9a59c25b512@rawbw.com> <1d688197-0ca7-025d-bdcd-4d23631182b2@gmail.com> <CALH631mMuVzGKw_wKrVTHbt-CeRsLKGiM7Yc-iBpAQ6QOSrRWw@mail.gmail.com> <20191028222852.mrgkywe5vjvsvpdk@atuin.in.mat.cc> <CALH631nr1B9y=7L-9LFbZBp900v-XTDPJCemhOFL-M=KQ76Kdw@mail.gmail.com> <20191030062749.tersvou4qmhg2xv2@atuin.in.mat.cc> <CALH631=jJqWH9UhTy0eXs1NVwbAk93U_FDh%2B6GN2Y98zJH_XLQ@mail.gmail.com> <20191030105537.GA18688@lonesome.com> <ecbed237-1f89-f39b-1b7e-43656ad5beda@gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 2019-10-30 05:42, Jason Bacon wrote:
>
> I can help clean it up, but not until the end of the year after I 
> finish some current projects.
>
>     JB
>

The problem with the science/py-tensorflow port stems from it being 
built by an unstable tool "bazel" that Google promotes. Bazel is 
overengineered, overly complex, hard to use, has obscure parts, and also 
changes frequently. It doesn't even have a feature to install files, 
like GNU Make or cmake. IMO, tensorflow can't just be "cleaned up" and 
built after this. Every Bazel-based port would suffer from the same 
problems. Google can afford to have such a tool, but other's don't have 
that kind of resources to handle these issues. I'd just fork the bazel 
port to the version that can build TensorFlow, this is the easiest solution.


Yuri





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?47ff19e3-b701-414f-c935-1c964c115f23>