Re: [gentoo-dev] Preventing $ARCH flags in USE
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Donnie Berkholz wrote: > On 09:19 Wed 17 Sep , Zac Medico wrote: >> I suggest that we unmask the appropriate ARCH flags in >> profiles/arch/*/use.mask, add ../base to profiles/arch/*/parent, and >> create profiles/arch/base/use.mask to mask all of the existing ARCH >> flags. This will serve to mask all but the appropriate ARCH flags >> for all of the 2008.0 profiles. Does this seem reasonable? > > Yes, and don't the arch profiles already have base as a parent? > They do now because I've already implemented the above suggestion. ;) - -- Thanks, Zac -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkjfsbIACgkQ/ejvha5XGaPhDgCfYltLRE/yAJYB09mugh2oTsvf 1BkAnjkvYlR2ohrjm5tnb8SbLxcDsgip =6kIP -END PGP SIGNATURE-
Re: [gentoo-dev] Preventing $ARCH flags in USE
On 09:19 Wed 17 Sep , Zac Medico wrote: > I suggest that we unmask the appropriate ARCH flags in > profiles/arch/*/use.mask, add ../base to profiles/arch/*/parent, and > create profiles/arch/base/use.mask to mask all of the existing ARCH > flags. This will serve to mask all but the appropriate ARCH flags > for all of the 2008.0 profiles. Does this seem reasonable? Yes, and don't the arch profiles already have base as a parent? -- Thanks, Donnie Donnie Berkholz Developer, Gentoo Linux Blog: http://dberkholz.wordpress.com pgphffCk99RNw.pgp Description: PGP signature
Re: [gentoo-dev] Preventing $ARCH flags in USE
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Vlastimil Babka wrote: > Apparently, setting USE=x86 in make.conf on amd64 arch can have funny > consequences such as [1]. And I can imagine even more subtle and hard to > detect errors due to this. > > I think it's better to prevent this rather than waste time with bug > reports like that. I asked Zac on IRC whether portage could filter such > flags. He suggested using use.mask in profiles. Well since ARCH is also > set by a profile, why not. Although a really persistent and stupid user > could use.unmask, it's better than no protection. And then we can think > how to replace the current ARCH->USE flag system with e.g. USE_EXPAND. > What do you think? > > Vlastimil > > [1] http://bugs.gentoo.org/show_bug.cgi?id=236801 > I suggest that we unmask the appropriate ARCH flags in profiles/arch/*/use.mask, add ../base to profiles/arch/*/parent, and create profiles/arch/base/use.mask to mask all of the existing ARCH flags. This will serve to mask all but the appropriate ARCH flags for all of the 2008.0 profiles. Does this seem reasonable? - -- Thanks, Zac -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkjRLiMACgkQ/ejvha5XGaOUnACfb+wsK5BbVdNgmuG/KShxDPXy hUUAn2a4hwO+4euOmExozx+7MegJZLK7 =9/4W -END PGP SIGNATURE-
Re: [gentoo-dev] Preventing $ARCH flags in USE
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Santiago M. Mola wrote: > On Mon, Sep 15, 2008 at 9:45 PM, Vlastimil Babka <[EMAIL PROTECTED]> wrote: >> I think it's better to prevent this rather than waste time with bug >> reports like that. I asked Zac on IRC whether portage could filter such >> flags. He suggested using use.mask in profiles. Well since ARCH is also >> set by a profile, why not. Although a really persistent and stupid user >> could use.unmask, it's better than no protection. And then we can think >> how to replace the current ARCH->USE flag system with e.g. USE_EXPAND. >> What do you think? >> > > Seems like an acceptable workaround. > > For future EAPIs, ARCH could be a regular USE_EXPANDed flag as you > suggest, and package managers could filter any flag in USE which is > not listed in IUSE. I suspect that it may be a little more than a "workaround". Consider a case where IUSE contains elibc_glibc and the current selected profile has set ELIBC=uclibc. In this case, the user could conceivable set USE=elibc_glibc in make.conf, which would clearly be an invalid setting. Therefore, it seems natural to mask the elibc_glibc USE flag on all profiles except those which actually use glibc. - -- Thanks, Zac -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkjRHTQACgkQ/ejvha5XGaPsBwCfQ1tv/AgKH4x0PS++QtbFeav0 3NAAoJbvO3FHjt3uGL/kffOxRh7/akZq =Ez2M -END PGP SIGNATURE-
Re: [gentoo-dev] Preventing $ARCH flags in USE
On 15-09-2008 13:20:23 -0700, Zac Medico wrote: > > For future EAPIs, ARCH could be a regular USE_EXPANDed flag as you > > suggest, and package managers could filter any flag in USE which is > > not listed in IUSE. > > While I don't necessarily disagree with you, my impression is that > most people tend to think that certain profile-specific flags such > as userland_* and kernel_* should be considered as implicit members > of IUSE and therefore they shouldn't be explicitly listed in in IUSE. Yes, IMO mainly because they should never explicitly be set by users, so they shouldn't get a hint they can set it either. -- Fabian Groffen Gentoo on a different level
Re: [gentoo-dev] Preventing $ARCH flags in USE
On Tue, 16 Sep 2008 08:01:46 +0200 Fabian Groffen <[EMAIL PROTECTED]> wrote: > > While I don't necessarily disagree with you, my impression is that > > most people tend to think that certain profile-specific flags such > > as userland_* and kernel_* should be considered as implicit members > > of IUSE and therefore they shouldn't be explicitly listed in in > > IUSE. > > Yes, IMO mainly because they should never explicitly be set by users, > so they shouldn't get a hint they can set it either. That's covered by USE_EXPAND_HIDDEN... -- Ciaran McCreesh signature.asc Description: PGP signature
Re: [gentoo-dev] Preventing $ARCH flags in USE
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Santiago M. Mola wrote: > On Mon, Sep 15, 2008 at 9:45 PM, Vlastimil Babka <[EMAIL PROTECTED]> wrote: >> I think it's better to prevent this rather than waste time with bug >> reports like that. I asked Zac on IRC whether portage could filter such >> flags. He suggested using use.mask in profiles. Well since ARCH is also >> set by a profile, why not. Although a really persistent and stupid user >> could use.unmask, it's better than no protection. And then we can think >> how to replace the current ARCH->USE flag system with e.g. USE_EXPAND. >> What do you think? >> > > Seems like an acceptable workaround. > > For future EAPIs, ARCH could be a regular USE_EXPANDed flag as you > suggest, and package managers could filter any flag in USE which is > not listed in IUSE. While I don't necessarily disagree with you, my impression is that most people tend to think that certain profile-specific flags such as userland_* and kernel_* should be considered as implicit members of IUSE and therefore they shouldn't be explicitly listed in in IUSE. - -- Thanks, Zac -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkjOw4YACgkQ/ejvha5XGaMILACfSQeRT7y1RzwTWWRnHvXBqFCh 9Q0An2AQZ9jJXLSWD1sKfL6+RdVNgEjl =Vh+a -END PGP SIGNATURE-
Re: [gentoo-dev] Preventing $ARCH flags in USE
On Mon, Sep 15, 2008 at 9:45 PM, Vlastimil Babka <[EMAIL PROTECTED]> wrote: > > I think it's better to prevent this rather than waste time with bug > reports like that. I asked Zac on IRC whether portage could filter such > flags. He suggested using use.mask in profiles. Well since ARCH is also > set by a profile, why not. Although a really persistent and stupid user > could use.unmask, it's better than no protection. And then we can think > how to replace the current ARCH->USE flag system with e.g. USE_EXPAND. > What do you think? > Seems like an acceptable workaround. For future EAPIs, ARCH could be a regular USE_EXPANDed flag as you suggest, and package managers could filter any flag in USE which is not listed in IUSE. Regards, -- Santiago M. Mola Jabber ID: [EMAIL PROTECTED]
[gentoo-dev] Preventing $ARCH flags in USE
Apparently, setting USE=x86 in make.conf on amd64 arch can have funny consequences such as [1]. And I can imagine even more subtle and hard to detect errors due to this. I think it's better to prevent this rather than waste time with bug reports like that. I asked Zac on IRC whether portage could filter such flags. He suggested using use.mask in profiles. Well since ARCH is also set by a profile, why not. Although a really persistent and stupid user could use.unmask, it's better than no protection. And then we can think how to replace the current ARCH->USE flag system with e.g. USE_EXPAND. What do you think? Vlastimil [1] http://bugs.gentoo.org/show_bug.cgi?id=236801 signature.asc Description: OpenPGP digital signature