Re: [devel] RPM %_tmppath - что плохого в силинках?

[Alexey Tourbin <at@altlinux.ru>]

[-- Attachment #1: Type: text/plain, Size: 3387 bytes --]

On Thu, Mar 13, 2008 at 09:50:59AM +0300, Хихин Руслан wrote:
> Здравствуйте Alexey Tourbin
>   В сообщении от 13 марта 2008 Alexey Tourbin написал(a):
>  > On Wed, Mar 12, 2008 at 01:22:10PM +0600, Ildar Mulyukov wrote:
>  > > find-lang: ERROR: non-canonical
>  > > Более того, почему с неканоническими путями без find-lang пакеты
>  > > собираются?
>  > Не собираются.
> А эту фичу нельзя отключить ?
> Всё равно окончательно rpm собирается в хешере, а вот до его сборки в 
> хешере иногда удобно его собирать в домашнем каталоге, причём 
> где-нибудь внутри /tmp, который на tmpfs.

Тогда придётся усложнять логику поиска зависимостей, причём не сразу
понятно в каких именно местах и чем это чревато.  А так её удаётся
упростить.

commit 810f4045edef59b4ba77521d2335a6c4713c357d
Author: Alexey Tourbin <at@altlinux>
Date:   Tue Feb 19 03:13:51 2008 +0300

    find-package: removed findpackage_xbroot stuff

diff --git a/scripts/find-package.in b/scripts/find-package.in
index 00b280a..d4c795b 100755
--- a/scripts/find-package.in
+++ b/scripts/find-package.in
@@ -58,12 +58,6 @@ Debug "RPM_FINDPACKAGE_ROOT_PATH=$RPM_FINDPACKAGE_ROOT_PATH"
 # Below we use 'local Verbose=Info' to increase per-case verbosity.
 Verbose=Verbose
 
-if [ -n "${RPM_BUILD_ROOT-}" ]; then
-	findpackage_xbroot=$(readlink -ve "$RPM_BUILD_ROOT")
-else
-	unset findpackage_xbroot ||:
-fi
-
 FindByPathUnderBuildroot()
 {
 	local f="$1" rep="$2" xbrep; shift 2 || return
@@ -72,12 +66,12 @@ FindByPathUnderBuildroot()
 	MatchBuildroot()
 	{
 		[ -n "$1" ] &&
-		[ -z "${1##$findpackage_xbroot/*}" ]
+		[ -z "${1##$RPM_BUILD_ROOT/*}" ]
 	}
 	StatPrint()
 	{
 		[ -e "$1" -o -L "$1" ] &&
-		printf '%s\n' "${1#$findpackage_xbroot}"
+		printf '%s\n' "${1#$RPM_BUILD_ROOT}"
 	}
 
 	xbrep=$(CanonPath "$RPM_BUILD_ROOT$rep")

commit e0598b6fe35fa42ef3737001d1ff38161c6bca84
Author: Alexey Tourbin <at@altlinux>
Date:   Tue Feb 19 02:38:42 2008 +0300

    functions (ValidateBuildRoot): require canonical RPM_BUILD_ROOT
    
    Some scripts like lib.req rely on the fact that RPM_BUILD_ROOT
    should not end with trailing slashes or something.  Other
    scripts like find-package explicitly assume that RPM_BUILD_ROOT
    can be relocated within filesystem; they use something like
    "real_buildroot" variables.  The code gets complicated,
    fragile, and error-prone.
    
    Therefore, guys, hereby I FORBID non-canonical RPM_BUILD_ROOT path.
    Note that RPM_BUILD_ROOT actually does not have to exist.
    This is another problem...

diff --git a/scripts/functions b/scripts/functions
index 9ffa7f0..636ef74 100644
--- a/scripts/functions
+++ b/scripts/functions
@@ -57,8 +57,16 @@ Debug()
 
 ValidateBuildRoot()
 {
+	[ -n "${RPM_BUILD_ROOT-}" ] ||
+		Fatal 'RPM_BUILD_ROOT not set'
 	[ -n "$(printf %s "$RPM_BUILD_ROOT" |tr -d ' /.')" ] ||
-		Fatal 'invalid $RPM_BUILD_ROOT'
+		Fatal "bogus RPM_BUILD_ROOT=$RPM_BUILD_ROOT"
+	local real_buildroot
+	# There could be simply no %install section...
+	# I pretend that non-existent buildroot is just canonical enough.
+	real_buildroot=$(readlink -vm -- "$RPM_BUILD_ROOT")
+	[ "$RPM_BUILD_ROOT" = "$real_buildroot" ] ||
+		Fatal "non-canonical RPM_BUILD_ROOT=$RPM_BUILD_ROOT real_buildroot=$real_buildroot"
 }
 
 [ -z "${RPM_BUILD_ROOT-}" ] || ValidateBuildRoot


[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]