feature flag change notifications

feature flag change notifications

[dann frazier]

Hi,
  I maintain the edk2 packages in the Debian and Ubuntu
distributions. A few times over the years I've had users report
regressions that turned out to be due to upstream build flag
changes. I wonder if it would be possible for upstream to communicate
such changes - either via release notes or, where possible, as
build-time checks - to us downstream projects. Here are the examples I
recall:

> commit 4de8d61bcec02a13ceed84f92b0cf3ea58adf9c5
> Author: Gerd Hoffmann <kraxel@redhat.com>
> Date:   Wed Dec 15 12:39:20 2021 +0100
> 
>     OvmfPkg: rework TPM configuration
> 

For this I needed to change -DTPM_ENABLE=TRUE to -DTPM2_ENABLE=TRUE. I
feel like a build-time check would've been a good way to communicate
this one, e.g.:

#ifdef TPM_ENABLE
#error TPM_ENABLE has been renamed to TPM2_ENABLE
#endif

> commit 1631bb26ae991e530d3c96fe3161ea15144b358e
> Author: Gary Lin <glin@suse.com>
> Date:   Mon Jun 10 14:55:09 2019 +0800
> 
>     OvmfPkg/README: Update the network build flags
>     
>     The following network build flags changed due to the inclusion of
>     NetworkPkg/Network.fdf.inc.
>
Similar to the above, I needed to deal with this by setting
-DNETWORK_HTTP_BOOT_ENABLE=TRUE instead of -DHTTP_BOOT_ENABLE=TRUE. It
seems like maybe an #error on the removed #define could've helped
here as well.

And finally:

> commit 57783adfb579da32b1eeda77b2bec028a5e0b7b3
> Author: Michael D Kinney <michael.d.kinney@intel.com>
> Date:   Tue Jul 26 12:40:00 2022 -0700
>
>     OvmfPkg: Change default to disable MptScsi and PvScsi

With this types of change, it would be useful just to know this is
coming and why so I can decide to either notify users of the change
and/or override the default to avoid the regression. Would it be
possible to describe such changes in the release notes at
https://github.com/tianocore/edk2/releases - and, if not urgent,
perhaps #warn of deprecation in the source for one stable release
cycle?

  -dann

Re: [edk2-devel] feature flag change notifications

[Michael D Kinney]

Hi Dann,

These are reasonable requests.

Adding Liming Gao who is currently responsible for the edk2 releases.

Let's start with release notes and see if some of these can be added
to the release documentation and add to the edk2 release process
to clearly note these type of changes in future release notes.

As far as #error and #warn.  Those are possible, but we would need
to do some analysis for the types of changes that would require
that behavior.  Your examples are useful as a starting point.

Once concern is how many of those #error/#warn conditions will
build up over time and then a process to remove them after an
expiration period.

Mike

> -----Original Message-----
> From: devel@edk2.groups.io <devel@edk2.groups.io> On Behalf Of dann frazier
> Sent: Monday, December 5, 2022 4:10 PM
> To: devel@edk2.groups.io
> Subject: [edk2-devel] feature flag change notifications
> 
> Hi,
>   I maintain the edk2 packages in the Debian and Ubuntu
> distributions. A few times over the years I've had users report
> regressions that turned out to be due to upstream build flag
> changes. I wonder if it would be possible for upstream to communicate
> such changes - either via release notes or, where possible, as
> build-time checks - to us downstream projects. Here are the examples I
> recall:
> 
> > commit 4de8d61bcec02a13ceed84f92b0cf3ea58adf9c5
> > Author: Gerd Hoffmann <kraxel@redhat.com>
> > Date:   Wed Dec 15 12:39:20 2021 +0100
> >
> >     OvmfPkg: rework TPM configuration
> >
> 
> For this I needed to change -DTPM_ENABLE=TRUE to -DTPM2_ENABLE=TRUE. I
> feel like a build-time check would've been a good way to communicate
> this one, e.g.:
> 
> #ifdef TPM_ENABLE
> #error TPM_ENABLE has been renamed to TPM2_ENABLE
> #endif
> 
> > commit 1631bb26ae991e530d3c96fe3161ea15144b358e
> > Author: Gary Lin <glin@suse.com>
> > Date:   Mon Jun 10 14:55:09 2019 +0800
> >
> >     OvmfPkg/README: Update the network build flags
> >
> >     The following network build flags changed due to the inclusion of
> >     NetworkPkg/Network.fdf.inc.
> >
> Similar to the above, I needed to deal with this by setting
> -DNETWORK_HTTP_BOOT_ENABLE=TRUE instead of -DHTTP_BOOT_ENABLE=TRUE. It
> seems like maybe an #error on the removed #define could've helped
> here as well.
> 
> And finally:
> 
> > commit 57783adfb579da32b1eeda77b2bec028a5e0b7b3
> > Author: Michael D Kinney <michael.d.kinney@intel.com>
> > Date:   Tue Jul 26 12:40:00 2022 -0700
> >
> >     OvmfPkg: Change default to disable MptScsi and PvScsi
> 
> With this types of change, it would be useful just to know this is
> coming and why so I can decide to either notify users of the change
> and/or override the default to avoid the regression. Would it be
> possible to describe such changes in the release notes at
> https://github.com/tianocore/edk2/releases - and, if not urgent,
> perhaps #warn of deprecation in the source for one stable release
> cycle?
> 
>   -dann
> 
> 
> 
> 

回复: [edk2-devel] feature flag change notifications

[gaoliming]

Dann:
  I base on new features to collect the incompatible changes, and add them into the updated notes. 

  If you find any missing, please send the mail to me. 

Thanks
Liming
> -----邮件原件-----
> 发件人: Kinney, Michael D <michael.d.kinney@intel.com>
> 发送时间: 2022年12月6日 9:22
> 收件人: devel@edk2.groups.io; dann.frazier@canonical.com; Gao, Liming
> <gaoliming@byosoft.com.cn>
> 主题: RE: [edk2-devel] feature flag change notifications
> 
> Hi Dann,
> 
> These are reasonable requests.
> 
> Adding Liming Gao who is currently responsible for the edk2 releases.
> 
> Let's start with release notes and see if some of these can be added
> to the release documentation and add to the edk2 release process
> to clearly note these type of changes in future release notes.
> 
> As far as #error and #warn.  Those are possible, but we would need
> to do some analysis for the types of changes that would require
> that behavior.  Your examples are useful as a starting point.
> 
> Once concern is how many of those #error/#warn conditions will
> build up over time and then a process to remove them after an
> expiration period.
> 
> Mike
> 
> > -----Original Message-----
> > From: devel@edk2.groups.io <devel@edk2.groups.io> On Behalf Of dann
> frazier
> > Sent: Monday, December 5, 2022 4:10 PM
> > To: devel@edk2.groups.io
> > Subject: [edk2-devel] feature flag change notifications
> >
> > Hi,
> >   I maintain the edk2 packages in the Debian and Ubuntu
> > distributions. A few times over the years I've had users report
> > regressions that turned out to be due to upstream build flag
> > changes. I wonder if it would be possible for upstream to communicate
> > such changes - either via release notes or, where possible, as
> > build-time checks - to us downstream projects. Here are the examples I
> > recall:
> >
> > > commit 4de8d61bcec02a13ceed84f92b0cf3ea58adf9c5
> > > Author: Gerd Hoffmann <kraxel@redhat.com>
> > > Date:   Wed Dec 15 12:39:20 2021 +0100
> > >
> > >     OvmfPkg: rework TPM configuration
> > >
> >
> > For this I needed to change -DTPM_ENABLE=TRUE to
> -DTPM2_ENABLE=TRUE. I
> > feel like a build-time check would've been a good way to communicate
> > this one, e.g.:
> >
> > #ifdef TPM_ENABLE
> > #error TPM_ENABLE has been renamed to TPM2_ENABLE
> > #endif
> >
> > > commit 1631bb26ae991e530d3c96fe3161ea15144b358e
> > > Author: Gary Lin <glin@suse.com>
> > > Date:   Mon Jun 10 14:55:09 2019 +0800
> > >
> > >     OvmfPkg/README: Update the network build flags
> > >
> > >     The following network build flags changed due to the inclusion of
> > >     NetworkPkg/Network.fdf.inc.
> > >
> > Similar to the above, I needed to deal with this by setting
> > -DNETWORK_HTTP_BOOT_ENABLE=TRUE instead of
> -DHTTP_BOOT_ENABLE=TRUE. It
> > seems like maybe an #error on the removed #define could've helped
> > here as well.
> >
> > And finally:
> >
> > > commit 57783adfb579da32b1eeda77b2bec028a5e0b7b3
> > > Author: Michael D Kinney <michael.d.kinney@intel.com>
> > > Date:   Tue Jul 26 12:40:00 2022 -0700
> > >
> > >     OvmfPkg: Change default to disable MptScsi and PvScsi
> >
> > With this types of change, it would be useful just to know this is
> > coming and why so I can decide to either notify users of the change
> > and/or override the default to avoid the regression. Would it be
> > possible to describe such changes in the release notes at
> > https://github.com/tianocore/edk2/releases - and, if not urgent,
> > perhaps #warn of deprecation in the source for one stable release
> > cycle?
> >
> >   -dann
> >
> >
> > 
> >

Re: [edk2-devel] feature flag change notifications

[dann frazier]

Hi Liming,

  Thank you, that is appreciated. Going forward I'll let you know if I
trip over any not already documented.

  -dann

On Tue, Dec 6, 2022 at 5:59 PM gaoliming via groups.io
<gaoliming=byosoft.com.cn@groups.io> wrote:
>
> Dann:
>   I base on new features to collect the incompatible changes, and add them into the updated notes.
>
>   If you find any missing, please send the mail to me.
>
> Thanks
> Liming
> > -----邮件原件-----
> > 发件人: Kinney, Michael D <michael.d.kinney@intel.com>
> > 发送时间: 2022年12月6日 9:22
> > 收件人: devel@edk2.groups.io; dann.frazier@canonical.com; Gao, Liming
> > <gaoliming@byosoft.com.cn>
> > 主题: RE: [edk2-devel] feature flag change notifications
> >
> > Hi Dann,
> >
> > These are reasonable requests.
> >
> > Adding Liming Gao who is currently responsible for the edk2 releases.
> >
> > Let's start with release notes and see if some of these can be added
> > to the release documentation and add to the edk2 release process
> > to clearly note these type of changes in future release notes.
> >
> > As far as #error and #warn.  Those are possible, but we would need
> > to do some analysis for the types of changes that would require
> > that behavior.  Your examples are useful as a starting point.
> >
> > Once concern is how many of those #error/#warn conditions will
> > build up over time and then a process to remove them after an
> > expiration period.
> >
> > Mike
> >
> > > -----Original Message-----
> > > From: devel@edk2.groups.io <devel@edk2.groups.io> On Behalf Of dann
> > frazier
> > > Sent: Monday, December 5, 2022 4:10 PM
> > > To: devel@edk2.groups.io
> > > Subject: [edk2-devel] feature flag change notifications
> > >
> > > Hi,
> > >   I maintain the edk2 packages in the Debian and Ubuntu
> > > distributions. A few times over the years I've had users report
> > > regressions that turned out to be due to upstream build flag
> > > changes. I wonder if it would be possible for upstream to communicate
> > > such changes - either via release notes or, where possible, as
> > > build-time checks - to us downstream projects. Here are the examples I
> > > recall:
> > >
> > > > commit 4de8d61bcec02a13ceed84f92b0cf3ea58adf9c5
> > > > Author: Gerd Hoffmann <kraxel@redhat.com>
> > > > Date:   Wed Dec 15 12:39:20 2021 +0100
> > > >
> > > >     OvmfPkg: rework TPM configuration
> > > >
> > >
> > > For this I needed to change -DTPM_ENABLE=TRUE to
> > -DTPM2_ENABLE=TRUE. I
> > > feel like a build-time check would've been a good way to communicate
> > > this one, e.g.:
> > >
> > > #ifdef TPM_ENABLE
> > > #error TPM_ENABLE has been renamed to TPM2_ENABLE
> > > #endif
> > >
> > > > commit 1631bb26ae991e530d3c96fe3161ea15144b358e
> > > > Author: Gary Lin <glin@suse.com>
> > > > Date:   Mon Jun 10 14:55:09 2019 +0800
> > > >
> > > >     OvmfPkg/README: Update the network build flags
> > > >
> > > >     The following network build flags changed due to the inclusion of
> > > >     NetworkPkg/Network.fdf.inc.
> > > >
> > > Similar to the above, I needed to deal with this by setting
> > > -DNETWORK_HTTP_BOOT_ENABLE=TRUE instead of
> > -DHTTP_BOOT_ENABLE=TRUE. It
> > > seems like maybe an #error on the removed #define could've helped
> > > here as well.
> > >
> > > And finally:
> > >
> > > > commit 57783adfb579da32b1eeda77b2bec028a5e0b7b3
> > > > Author: Michael D Kinney <michael.d.kinney@intel.com>
> > > > Date:   Tue Jul 26 12:40:00 2022 -0700
> > > >
> > > >     OvmfPkg: Change default to disable MptScsi and PvScsi
> > >
> > > With this types of change, it would be useful just to know this is
> > > coming and why so I can decide to either notify users of the change
> > > and/or override the default to avoid the regression. Would it be
> > > possible to describe such changes in the release notes at
> > > https://github.com/tianocore/edk2/releases - and, if not urgent,
> > > perhaps #warn of deprecation in the source for one stable release
> > > cycle?
> > >
> > >   -dann
> > >
> > >
> > >
> > >
>
>
>
>
>
> 
>
>

Re: [edk2-devel] feature flag change notifications

[dann frazier]

Hi Mike,

  Thanks for the consideration. I think the release notes are a
sufficient starting point.

  -dann

On Mon, Dec 5, 2022 at 6:22 PM Kinney, Michael D
<michael.d.kinney@intel.com> wrote:
>
> Hi Dann,
>
> These are reasonable requests.
>
> Adding Liming Gao who is currently responsible for the edk2 releases.
>
> Let's start with release notes and see if some of these can be added
> to the release documentation and add to the edk2 release process
> to clearly note these type of changes in future release notes.
>
> As far as #error and #warn.  Those are possible, but we would need
> to do some analysis for the types of changes that would require
> that behavior.  Your examples are useful as a starting point.
>
> Once concern is how many of those #error/#warn conditions will
> build up over time and then a process to remove them after an
> expiration period.
>
> Mike
>
> > -----Original Message-----
> > From: devel@edk2.groups.io <devel@edk2.groups.io> On Behalf Of dann frazier
> > Sent: Monday, December 5, 2022 4:10 PM
> > To: devel@edk2.groups.io
> > Subject: [edk2-devel] feature flag change notifications
> >
> > Hi,
> >   I maintain the edk2 packages in the Debian and Ubuntu
> > distributions. A few times over the years I've had users report
> > regressions that turned out to be due to upstream build flag
> > changes. I wonder if it would be possible for upstream to communicate
> > such changes - either via release notes or, where possible, as
> > build-time checks - to us downstream projects. Here are the examples I
> > recall:
> >
> > > commit 4de8d61bcec02a13ceed84f92b0cf3ea58adf9c5
> > > Author: Gerd Hoffmann <kraxel@redhat.com>
> > > Date:   Wed Dec 15 12:39:20 2021 +0100
> > >
> > >     OvmfPkg: rework TPM configuration
> > >
> >
> > For this I needed to change -DTPM_ENABLE=TRUE to -DTPM2_ENABLE=TRUE. I
> > feel like a build-time check would've been a good way to communicate
> > this one, e.g.:
> >
> > #ifdef TPM_ENABLE
> > #error TPM_ENABLE has been renamed to TPM2_ENABLE
> > #endif
> >
> > > commit 1631bb26ae991e530d3c96fe3161ea15144b358e
> > > Author: Gary Lin <glin@suse.com>
> > > Date:   Mon Jun 10 14:55:09 2019 +0800
> > >
> > >     OvmfPkg/README: Update the network build flags
> > >
> > >     The following network build flags changed due to the inclusion of
> > >     NetworkPkg/Network.fdf.inc.
> > >
> > Similar to the above, I needed to deal with this by setting
> > -DNETWORK_HTTP_BOOT_ENABLE=TRUE instead of -DHTTP_BOOT_ENABLE=TRUE. It
> > seems like maybe an #error on the removed #define could've helped
> > here as well.
> >
> > And finally:
> >
> > > commit 57783adfb579da32b1eeda77b2bec028a5e0b7b3
> > > Author: Michael D Kinney <michael.d.kinney@intel.com>
> > > Date:   Tue Jul 26 12:40:00 2022 -0700
> > >
> > >     OvmfPkg: Change default to disable MptScsi and PvScsi
> >
> > With this types of change, it would be useful just to know this is
> > coming and why so I can decide to either notify users of the change
> > and/or override the default to avoid the regression. Would it be
> > possible to describe such changes in the release notes at
> > https://github.com/tianocore/edk2/releases - and, if not urgent,
> > perhaps #warn of deprecation in the source for one stable release
> > cycle?
> >
> >   -dann
> >
> >
> > 
> >
>