[gentoo-dev] epatch: splitting out common options from user-specific ones

[gentoo-dev] epatch: splitting out common options from user-specific ones

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 861 bytes --]

it isn't uncommon for people to want to force the patch (-p#) or fuzz (-f#) 
level when applying specific patches.  but it is unusual that they want to kill 
off the extra options: -g0 -E --no-backup-if-mismatch.  so i'd like to split 
these off and improve the epatch API.

# Extra options to pass to `patch` (such as -p1).
EPATCH_OPTS=""
# Common options to pass to `patch` (you probably shouldn't need
# to change these at all).
EPATCH_COMMON_OPTS="-g0 -E --no-backup-if-mismatch"

and then i'll try to extend `epatch` so the first set of arguments can 
implicitly set EPATCH_OPTS for that one patch:
	epatch -p1 "${FILESDIR}"/${P}-foo.patch
this is much nicer than the current:
	EPATCH_OPTS="-p1" epatch "${FILESDIR}"/${P}-foo.patch

i can't see this causing any issues, but considering everyone uses `epatch`, 
best to ask first.
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [gentoo-dev] epatch: splitting out common options from user-specific ones

[Mike Frysinger]

[-- Attachment #1: Type: Text/Plain, Size: 3678 bytes --]

no complaints, so here's the patch.  precedence order is EPATCH_COMMON_OPTS
then EPATCH_OPTS then whatever has been specified on the cmdline.

so you can do:
	EPATCH_OPTS="-F0"
	epatch <all these patches will use -F0>
	epatch -p0 <patches will use -p0 and -F0>
	epatch <all these patches will use -F0>
(more for highlighting precedence than a realistic use case)
-mike

--- eutils.eclass
+++ eutils.eclass
@@ -230,13 +230,21 @@
 EPATCH_SUFFIX="patch.bz2"
 # @VARIABLE: EPATCH_OPTS
 # @DESCRIPTION:
-# Default options for patch:
+# Options to pass to patch.  Meant for ebuild/package-specific tweaking
+# such as forcing the patch level (-p#) or fuzz (-F#) factor.  Note that
+# for single patch tweaking, you can also pass flags directly to epatch.
+EPATCH_OPTS=""
+# @VARIABLE: EPATCH_COMMON_OPTS
+# @DESCRIPTION:
+# Common options to pass to `patch`.  You probably should never need to
+# change these.  If you do, please discuss it with base-system first to
+# be sure.
 # @CODE
 #	-g0 - keep RCS, ClearCase, Perforce and SCCS happy #24571
 #	--no-backup-if-mismatch - do not leave .orig files behind
 #	-E - automatically remove empty files
 # @CODE
-EPATCH_OPTS="-g0 -E --no-backup-if-mismatch"
+EPATCH_COMMON_OPTS="-g0 -E --no-backup-if-mismatch"
 # @VARIABLE: EPATCH_EXCLUDE
 # @DESCRIPTION:
 # List of patches not to apply.	 Note this is only file names,
@@ -257,7 +265,7 @@ EPATCH_MULTI_MSG="Applying various patch
 EPATCH_FORCE="no"
 
 # @FUNCTION: epatch
-# @USAGE: [patches] [dirs of patches]
+# @USAGE: [options] [patches] [dirs of patches]
 # @DESCRIPTION:
 # epatch is designed to greatly simplify the application of patches.  It can
 # process patch files directly, or directories of patches.  The patches may be
@@ -265,8 +273,12 @@ EPATCH_FORCE="no"
 # the -p option as epatch will automatically attempt -p0 to -p5 until things
 # apply successfully.
 #
-# If you do not specify any options, then epatch will default to the directory
-# specified by EPATCH_SOURCE.
+# If you do not specify any patches/dirs, then epatch will default to the
+# directory specified by EPATCH_SOURCE.
+#
+# Any options specified that start with a dash will be passed down to patch
+# for this specific invocation.  As soon as an arg w/out a dash is found, then
+# arg processing stops.
 #
 # When processing directories, epatch will apply all patches that match:
 # @CODE
@@ -294,6 +306,18 @@ epatch() {
 
 	unset P4CONFIG P4PORT P4USER # keep perforce at bay #56402
 
+	# First process options.  We localize the EPATCH_OPTS setting
+	# from above so that we can pass it on in the loop below with
+	# any additional values the user has specified.
+	local EPATCH_OPTS=( ${EPATCH_OPTS[*]} )
+	while [[ $# -gt 0 ]] ; do
+		case $1 in
+		-*) EPATCH_OPTS+=( "$1" ) ;;
+		*) break ;;
+		esac
+		shift
+	done
+
 	# Let the rest of the code process one user arg at a time --
 	# each arg may expand into multiple patches, and each arg may
 	# need to start off with the default global EPATCH_xxx values
@@ -305,6 +329,10 @@ epatch() {
 		return 0
 	fi
 
+	# Now that we know we're actually going to apply something, merge
+	# all of the patch options back in to a single variable for below.
+	EPATCH_OPTS="${EPATCH_COMMON_OPTS} ${EPATCH_OPTS[*]}"
+
 	local SINGLE_PATCH="no"
 	# no args means process ${EPATCH_SOURCE}
 	[[ $# -eq 0 ]] && set -- "${EPATCH_SOURCE}"
@@ -445,6 +473,7 @@ epatch() {
 		local patch_cmd
 		while [[ ${count} -lt 5 ]] ; do
 			patch_cmd="${BASH_ALIASES[patch]:-patch} -p${count} ${EPATCH_OPTS}"
+einfo $patch_cmd
 
 			# Generate some useful debug info ...
 			(

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [gentoo-dev] epatch: splitting out common options from user-specific ones

[Mike Frysinger]

[-- Attachment #1: Type: Text/Plain, Size: 338 bytes --]

On Thursday 19 April 2012 23:24:19 Mike Frysinger wrote:
> @@ -445,6 +473,7 @@ epatch() {
>  		local patch_cmd
>  		while [[ ${count} -lt 5 ]] ; do
>  			patch_cmd="${BASH_ALIASES[patch]:-patch} -p${count} 
> +einfo $patch_cmd
> 
>  			# Generate some useful debug info ...
>  			(

err, meant to delete that debug line
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [gentoo-dev] epatch: splitting out common options from user-specific ones

[Leho Kraav]

On Wednesday, April 18, 2012 9:10:02 PM UTC+3, Mike Frysinger wrote:
> it isn't uncommon for people to want to force the patch (-p#) or fuzz (-f#) 
> level when applying specific patches.  but it is unusual that they want to kill 
> off the extra options: -g0 -E --no-backup-if-mismatch.  so i'd like to split 
> these off and improve the epatch API.

LOL mike, i'm fighting with this thing at the very moment. there's a patch i received that has whitespace issues and i'd like to specify custom EPATCH_OPTS.

i applied the patch to my eclass overlay, and einfo is showing that the new eclass is used.

$ EPATCH_OPTS="--ignore-whitespace" emerge -va pf-sources

isn't achieved though. do you mean "cmdline" as in only within ebuild? what's the optimal way here? i was desperately searching for PATCHCOMMAND in "man make.conf".

Re: [gentoo-dev] epatch: splitting out common options from user-specific ones

[Leho Kraav]

On Wednesday, April 18, 2012 9:10:02 PM UTC+3, Mike Frysinger wrote:
> it isn't uncommon for people to want to force the patch (-p#) or fuzz (-f#)
> level when applying specific patches.  but it is unusual that they want to kill
> off the extra options: -g0 -E --no-backup-if-mismatch.  so i'd like to split
> these off and improve the epatch API.

LOL mike, i'm fighting with this thing at the very moment. there's a 
patch i received that has whitespace issues and i'd like to specify 
custom EPATCH_OPTS.

i applied the patch to my eclass overlay, and einfo is showing that the 
new eclass is used.

$ EPATCH_OPTS="--ignore-whitespace" emerge -va pf-sources

isn't achieved though. if i do:

236 EPATCH_OPTS="${EPATCH_OPTS}"

everything works as expected. do you mean "cmdline" as in only within 
ebuild? what's the optimal way here? i was desperately searching for 
PATCHCOMMAND in "man make.conf", but this definitely would do the job 
for me.

Re: [gentoo-dev] epatch: splitting out common options from user-specific ones

[Mike Frysinger]

[-- Attachment #1: Type: Text/Plain, Size: 903 bytes --]

On Friday 20 April 2012 15:38:19 Leho Kraav wrote:
> On Wednesday, April 18, 2012 9:10:02 PM UTC+3, Mike Frysinger wrote:
> > it isn't uncommon for people to want to force the patch (-p#) or fuzz
> > (-f#) level when applying specific patches.  but it is unusual that they
> > want to kill off the extra options: -g0 -E --no-backup-if-mismatch.  so
> > i'd like to split these off and improve the epatch API.
> 
> LOL mike, i'm fighting with this thing at the very moment. there's a patch
> i received that has whitespace issues and i'd like to specify custom
> EPATCH_OPTS.
> 
> i applied the patch to my eclass overlay, and einfo is showing that the new
> eclass is used.

i just committed it since no one responded.  so sync up.

> $ EPATCH_OPTS="--ignore-whitespace" emerge -va pf-sources

that's not the intention.  EPATCH_OPTS modification should be inside the 
ebuild.
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [gentoo-dev] epatch: splitting out common options from user-specific ones

[Leho Kraav]

On 20.04.2012 22:59, Mike Frysinger wrote:
> i just committed it since no one responded.  so sync up.
>
>> $ EPATCH_OPTS="--ignore-whitespace" emerge -va pf-sources
>
> that's not the intention.  EPATCH_OPTS modification should be inside the
> ebuild.

I can't see how having this available could hurt anything. Am I blind?

Re: [gentoo-dev] epatch: splitting out common options from user-specific ones

[Mike Frysinger]

[-- Attachment #1: Type: Text/Plain, Size: 569 bytes --]

On Friday 20 April 2012 16:06:02 Leho Kraav wrote:
> On 20.04.2012 22:59, Mike Frysinger wrote:
> > i just committed it since no one responded.  so sync up.
> > 
> >> $ EPATCH_OPTS="--ignore-whitespace" emerge -va pf-sources
> > 
> > that's not the intention.  EPATCH_OPTS modification should be inside the
> > ebuild.
> 
> I can't see how having this available could hurt anything. Am I blind?

i could remove EPATCH_OPTS="" from the eclass, but you'd get inconsistent 
behavior.  many ebuilds atm set that, so you wouldn't be able to override it.
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]