From: Ciaran McCreesh <ciaran.mccreesh@googlemail.com>
To: Christian Faulhammer <fauli@gentoo.org>
Cc: gentoo-pms@lists.gentoo.org
Subject: Re: [gentoo-pms] Patches for your reviewing pleasure: Various improvements, including cheat sheet
Date: Tue, 14 Apr 2009 15:46:57 +0100 [thread overview]
Message-ID: <20090414154657.37fb5ad8@snowcone> (raw)
In-Reply-To: <20090414155543.16e78e66@terra.solaris>
[-- Attachment #1: Type: text/plain, Size: 5488 bytes --]
On Tue, 14 Apr 2009 15:55:43 +0200
Christian Faulhammer <fauli@gentoo.org> wrote:
> > I agree with you on the principle, but not on the numbers. It's not
> > worth using subdirectories until 'ls' output won't fit on one
> > screen.
>
> Having actually done the change, does make the thing look really
> good. After a make run all the .log, .aux and whatever files are also
> around.
Well, it's really an all or nothing thing. And all's going to be a pain
until we're sure we aren't going to have to go back and edit the
history.
> 0001: Desk reference, updated and all issues you raised handled.
You're indenting with two spaces rather than four, and don't have our
precious vim modelines. Lack of consistency makes me cry!
> + from the same checkout as this overview.
has trailing whitespace. Git moans about this.
> + Please report mistakes or enhancements to this document via the
should read "mistakes in or enhancements to"
> +Currently there is no full specification what EAPI 0 includes.
> +Portage, official ebuild documentation and existing ebuilds set the
> +standard here, although obvious errors are still errors.
I'm wondering whether we can tone this (and the similar wording in PMS)
down a bit. The Council approved PMS as "correct, except where people
file bugs" a while back.
> + disabled explicitely by user configuration) with a \code{+} sign in
Spelling of 'explicitly'.
> +\item[Named slot dependencies] Dependencies can explicitely request a
Ditto.
> +\item[\code{SRC\_URI} arrows] Allows redirection of upstream file
> + naming scheme by using an "arrow" (\code{->}) to a new file name in
> + DISTDIR. See \featureref{src-uri-arrows}.
Not sure on this wording. How about something like:
SRC_URI="http://some/url -> foo" can be used to use foo as the
filename for downloads.
> +\item[USE dependencies] Dependencies can be specified to be built with
> + or without a specific USE flag, thus getting rid of countless
> + \code{built\_with\_use} checks. A more sophisticated syntax is
> + available in EAPI 3.
Dependencies can specify use flag requirements on their target,
removing the need for built_with_use checks. A more powerful syntax
that does not require the flag to be in IUSE is in EAPI 3.
> + Patching and other preparing tasks must no longer take place in
> + \code{src\_unpack} but in \code{src\_prepare}.
Patching and similar preparation must now be done in src_prepare,
not src_unpack.
> + be called via \code{default\_\emph{phasename}}, so small additions
> + you need will not be accompanied by a complete reimplementation of
> + the phase. Additionally as a shorthand for the current phase's
> + default definition \code{default} can be invoked.
so duplicating the standard implementation is no longer necessary.
The default function calls the current phase's default_ function
automatically.
> + Duplicating checks from the \code{pkg\_setup} phase may be
> + necessary.
Most checks should usually be repeated in pkg_setup.
> + phase can be called even for non-installed packages. Do not depend
> + on any outside programs in that case! See \featureref{pkg-info}.
Be warned that dependencies might not have been installed.
> +\item[Slot operators] There are two ways to define a wanted slot in a
> + dependency string:
Dependencies that are both DEPEND and RDEPEND and that can match
multiple slots should specify one of:
> +\item[USE dependency defaults] Additionally to the features offered in
In addition to the ...
> + after a USE flag (mind the braces).
Braces are {}. You want brackets or parentheses.
> The former leads to the
> + assumption that the USE flag is available when not found in the
> + dependency, the latter leads to the reverse behaviour. This mimicks
> + parts of the behaviour of \code{-{}-missing} in
The former specifies that flags not in IUSE should be treated as
enabled; the latter, disabled. Cannot be used with USE_EXPAND flags.
> + \code{/usr/share\{doc,info,man\}} are compressed on-disk after
s/are/may be
> + With \code{docompress} all arguments are passed to the inclusion
> + list, while the behaviour can be inverted with the \code{-x}
> + switch. See \featureref{controllable-compress}.
docompress path ... adds paths to the inclusion list for
compression. docompress -x path ... adds paths to the exclusion
list.
> + files respectively in the known manner.
"in the known manner" sounds weird.
> + command given as argument will not abort the build process in case
> + of a failure but preserve the exit status nonetheless. See
I'd mention that commands die on error before this. And then:
will return non-zero on failure rather than aborting the build.
> +\item[USE flags in any-of] Inside of \code{||} constructs USE flag
> + conditionals are forbidden (with exceptions).
use? ( ) blocks are forbidden as direct children of || ( ) blocks.
> + anymore.
'anymore' as a single word is generally considered horrible.
> 0002: Some Makefile tweaking (really minor).
OK, but won't apply easily without 0001, and I'm lazy.
> 0003: Use plainurl.bst from the system.
Applied, thanks.
> 0004: Some proper using of the description environment.
Applied, thanks.
--
Ciaran McCreesh
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]
next prev parent reply other threads:[~2009-04-14 14:47 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-04-12 19:14 [gentoo-pms] Patches for your reviewing pleasure: Various improvements, including cheat sheet Christian Faulhammer
2009-04-12 19:36 ` Ciaran McCreesh
2009-04-13 9:01 ` Christian Faulhammer
2009-04-13 16:01 ` Ciaran McCreesh
2009-04-14 8:44 ` Christian Faulhammer
2009-04-14 13:04 ` Ciaran McCreesh
2009-04-14 13:55 ` Christian Faulhammer
2009-04-14 14:46 ` Ciaran McCreesh [this message]
2009-04-15 15:28 ` Christian Faulhammer
2009-04-15 20:35 ` Ciaran McCreesh
2009-04-15 21:58 ` Christian Faulhammer
2009-04-15 22:10 ` Ciaran McCreesh
2009-04-15 22:46 ` Christian Faulhammer
2009-04-16 14:16 ` Ciaran McCreesh
2009-04-16 22:23 ` Christian Faulhammer
2009-04-17 14:11 ` Yannick Mortier
2009-04-12 21:25 ` David Leverton
2009-04-13 9:05 ` Christian Faulhammer
2009-04-13 21:16 ` [gentoo-pms] [PATCH] Work around TeX4ht being weird David Leverton
2009-04-14 8:32 ` Christian Faulhammer
2009-04-15 19:03 ` David Leverton
2009-04-15 21:59 ` Christian Faulhammer
2009-04-16 7:25 ` David Leverton
2009-04-16 19:37 ` [gentoo-pms] [PATCH 1/2] " David Leverton
2009-04-16 19:37 ` [gentoo-pms] [PATCH 2/2] Update TeX4ht sed hacks David Leverton
2009-04-16 19:42 ` Ciaran McCreesh
2009-04-17 14:13 ` Yannick Mortier
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20090414154657.37fb5ad8@snowcone \
--to=ciaran.mccreesh@googlemail.com \
--cc=fauli@gentoo.org \
--cc=gentoo-pms@lists.gentoo.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox