Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-09 Thread Sudip Mukherjee
On Fri, Feb 7, 2020 at 7:58 PM Sudip Mukherjee
 wrote:
>
> On Fri, Feb 7, 2020 at 10:17 AM Sudip Mukherjee
>  wrote:
> >
> > On Thu, Feb 6, 2020 at 10:53 PM Christian Barcenas
> >  wrote:
> > >
> > > I just noticed that your packaging repo is currently empty.
> > > Would you be able to push your current progress to Github
> > > so that it's easier to review the source package?
> >
> > Pushed now. Its without any epoch in the version. I will add the epoch
> > and push again tonight after back from $dayjob.
>
> Also pushed in wip/epoch branch with the epoch in only binary packages
> for your review. I have not yet uploaded to mentors.

And I have now pushed to mentors the package with epoch only in the
binary packages and I hope that is the consensus here.


-- 
Regards
Sudip



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-07 Thread Sudip Mukherjee
On Fri, Feb 7, 2020 at 10:17 AM Sudip Mukherjee
 wrote:
>
> On Thu, Feb 6, 2020 at 10:53 PM Christian Barcenas
>  wrote:
> >
> > I just noticed that your packaging repo is currently empty.
> > Would you be able to push your current progress to Github
> > so that it's easier to review the source package?
>
> Pushed now. Its without any epoch in the version. I will add the epoch
> and push again tonight after back from $dayjob.

Also pushed in wip/epoch branch with the epoch in only binary packages
for your review. I have not yet uploaded to mentors.

-- 
Regards
Sudip



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-07 Thread Guillem Jover
On Thu, 2020-02-06 at 09:39:29 +0100, Michael Biebl wrote:
> Am 06.02.20 um 09:22 schrieb Adam D. Barratt:
> > On 2020-02-06 08:12, Paul Gevers wrote:
> > > But, if I am correct, the source could be using a version without epoch
> > > and only use the epoch in the binary package (which can be dropped if
> > > libbpf0 is ever replaced by libbpf1).
> > 
> > Yes. It's a little more work on the packaging side, but entirely
> > possible to do it that way.

While this case is precisely what epochs were designed for, I'd still
try to minimize its usage here as much as possible. And go for no
epoch on the source packages, and prefixing the epoch in the binary
packages, because the one in libbpf1 could then be dropped.

> There is also libbpf-dev, which could not drop the epoch on a soname
> bump. Would be kinda odd if going forward libbpfx would not have an
> epoch and libbpf-dev does.

We already have similar cases in the archive. And the nice thing of
not going with a full epoch bump is that it can always be bumped for
the source later on, while the reverse is not possible. :)

> One other alternative could be, to ask your upstream to change the
> versioning scheme and instead of using 0.0.6, switch to 6.0.0 as first
> version number (which would be higher then 5.x)
> Other distros might have similar problems.

That might be even better, yes. :)

Thanks,
Guillem



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-07 Thread Sudip Mukherjee
On Thu, Feb 6, 2020 at 10:53 PM Christian Barcenas
 wrote:
>
> I just noticed that your packaging repo is currently empty.
> Would you be able to push your current progress to Github
> so that it's easier to review the source package?

Pushed now. Its without any epoch in the version. I will add the epoch
and push again tonight after back from $dayjob.


-- 
Regards
Sudip



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-06 Thread Christian Barcenas
I just noticed that your packaging repo is currently empty.
Would you be able to push your current progress to Github
so that it's easier to review the source package?

On Thu, Feb 6, 2020 at 2:34 PM Sudip Mukherjee
 wrote:
> So, do we also use epoch or shall I try the way which Paul suggested
> to add epoch only to the binary package?

My vote is to use libbpf's new versioning scheme and epoch 1 for both
source and binary packages. The primary benefit is simplicity:
IMO it's easier to reason about things that way.

It looks like upstream is trying to separate their own release
schedule from the kernel's. There is now a possibility that some
future libbpf release may contain changes not present in the
upstream kernel tree at all.

In this scenario, it would *only* make sense to use libbpf versioning
for the source package, so we might as well get both epoch adjustments
out of the way right now instead of raising this question again in the future.

Christian



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-06 Thread Sudip Mukherjee
On Thu, Feb 6, 2020 at 11:07 AM Sudip Mukherjee
 wrote:
>
> On Thu, Feb 6, 2020 at 8:39 AM Michael Biebl  wrote:
> >
> > Am 06.02.20 um 09:22 schrieb Adam D. Barratt:
> > > On 2020-02-06 08:12, Paul Gevers wrote:
> > >> Hi,
> > >>
> > >> On 06-02-2020 00:07, Adam D. Barratt wrote:
> > >>> On Wed, 2020-02-05 at 22:42 +, Sudip Mukherjee wrote:
> >  On Wed, Feb 5, 2020 at 10:22 PM Christian Barcenas
> >   wrote:
> > > Because this changes the versioning scheme from kernel releases
> > > (libbpf-dev and libbpf0 currently are at 5.4.13-1 in sid) to libbpf
> > > version numbers (0.0.6-1), the epoch needs to be incremented to 1 I
> > > believe.
> > 
> >  I had this doubt but 5.4.13-1 is the linux source version, and
> >  libbpf0 has the version of 0.0.5. And since there is no separate
> >  source for libbpf so will this not be  considered as a new package
> >  rather than a version change?
> > >>>
> 
> >
> > One other alternative could be, to ask your upstream to change the
> > versioning scheme and instead of using 0.0.6, switch to 6.0.0 as first
> > version number (which would be higher then 5.x)
> > Other distros might have similar problems.
>
> I think other distros already had the package split and are now using
> the source from github. Atleast Fedora and Gentoo have done already.
> Looks like Fedora solved the problem using epoch as can be seen in the
> comment at 
> https://src.fedoraproject.org/rpms/libbpf/blob/f30/f/libbpf.spec#_15

And, I have confirmed from Jiri who is the Fedora maintainer for
libbpf that they have used epoch to solve the version problem.
So, do we also use epoch or shall I try the way which Paul suggested
to add epoch only to the binary package?


-- 
Regards
Sudip



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-06 Thread Sudip Mukherjee
On Thu, Feb 6, 2020 at 8:39 AM Michael Biebl  wrote:
>
> Am 06.02.20 um 09:22 schrieb Adam D. Barratt:
> > On 2020-02-06 08:12, Paul Gevers wrote:
> >> Hi,
> >>
> >> On 06-02-2020 00:07, Adam D. Barratt wrote:
> >>> On Wed, 2020-02-05 at 22:42 +, Sudip Mukherjee wrote:
>  On Wed, Feb 5, 2020 at 10:22 PM Christian Barcenas
>   wrote:
> > Because this changes the versioning scheme from kernel releases
> > (libbpf-dev and libbpf0 currently are at 5.4.13-1 in sid) to libbpf
> > version numbers (0.0.6-1), the epoch needs to be incremented to 1 I
> > believe.
> 
>  I had this doubt but 5.4.13-1 is the linux source version, and
>  libbpf0 has the version of 0.0.5. And since there is no separate
>  source for libbpf so will this not be  considered as a new package
>  rather than a version change?
> >>>

>
> One other alternative could be, to ask your upstream to change the
> versioning scheme and instead of using 0.0.6, switch to 6.0.0 as first
> version number (which would be higher then 5.x)
> Other distros might have similar problems.

I think other distros already had the package split and are now using
the source from github. Atleast Fedora and Gentoo have done already.
Looks like Fedora solved the problem using epoch as can be seen in the
comment at https://src.fedoraproject.org/rpms/libbpf/blob/f30/f/libbpf.spec#_15


-- 
Regards
Sudip



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-06 Thread Michael Biebl
Am 06.02.20 um 09:22 schrieb Adam D. Barratt:
> On 2020-02-06 08:12, Paul Gevers wrote:
>> Hi,
>>
>> On 06-02-2020 00:07, Adam D. Barratt wrote:
>>> On Wed, 2020-02-05 at 22:42 +, Sudip Mukherjee wrote:
 On Wed, Feb 5, 2020 at 10:22 PM Christian Barcenas
  wrote:
> Because this changes the versioning scheme from kernel releases
> (libbpf-dev and libbpf0 currently are at 5.4.13-1 in sid) to libbpf
> version numbers (0.0.6-1), the epoch needs to be incremented to 1 I
> believe.

 I had this doubt but 5.4.13-1 is the linux source version, and
 libbpf0 has the version of 0.0.5. And since there is no separate
 source for libbpf so will this not beĀ  considered as a new package
 rather than a version change?
>>>
>>> No. There is currently a libbpf0 binary package. After the source
>>> package split, there will still be a libbpf0 binary package. The
>>> version of that binary package cannot go backwards.
>>>
>>> (The source package will indeed be NEW, but that's not the issue under
>>> discussion here.)
>>
>> But, if I am correct, the source could be using a version without epoch
>> and only use the epoch in the binary package (which can be dropped if
>> libbpf0 is ever replaced by libbpf1).
> 
> Yes. It's a little more work on the packaging side, but entirely
> possible to do it that way.

There is also libbpf-dev, which could not drop the epoch on a soname
bump. Would be kinda odd if going forward libbpfx would not have an
epoch and libbpf-dev does.

One other alternative could be, to ask your upstream to change the
versioning scheme and instead of using 0.0.6, switch to 6.0.0 as first
version number (which would be higher then 5.x)
Other distros might have similar problems.

Michael




signature.asc
Description: OpenPGP digital signature


Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-06 Thread Adam D. Barratt

On 2020-02-06 08:12, Paul Gevers wrote:

Hi,

On 06-02-2020 00:07, Adam D. Barratt wrote:

On Wed, 2020-02-05 at 22:42 +, Sudip Mukherjee wrote:

On Wed, Feb 5, 2020 at 10:22 PM Christian Barcenas
 wrote:

Because this changes the versioning scheme from kernel releases
(libbpf-dev and libbpf0 currently are at 5.4.13-1 in sid) to libbpf
version numbers (0.0.6-1), the epoch needs to be incremented to 1 I
believe.


I had this doubt but 5.4.13-1 is the linux source version, and
libbpf0 has the version of 0.0.5. And since there is no separate
source for libbpf so will this not be  considered as a new package
rather than a version change?


No. There is currently a libbpf0 binary package. After the source
package split, there will still be a libbpf0 binary package. The
version of that binary package cannot go backwards.

(The source package will indeed be NEW, but that's not the issue under
discussion here.)


But, if I am correct, the source could be using a version without epoch
and only use the epoch in the binary package (which can be dropped if
libbpf0 is ever replaced by libbpf1).


Yes. It's a little more work on the packaging side, but entirely 
possible to do it that way.


Regards,

Adam



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-06 Thread Paul Gevers
Hi,

On 06-02-2020 00:07, Adam D. Barratt wrote:
> On Wed, 2020-02-05 at 22:42 +, Sudip Mukherjee wrote:
>> On Wed, Feb 5, 2020 at 10:22 PM Christian Barcenas
>>  wrote:
>>> Because this changes the versioning scheme from kernel releases
>>> (libbpf-dev and libbpf0 currently are at 5.4.13-1 in sid) to libbpf
>>> version numbers (0.0.6-1), the epoch needs to be incremented to 1 I
>>> believe.
>>
>> I had this doubt but 5.4.13-1 is the linux source version, and
>> libbpf0 has the version of 0.0.5. And since there is no separate
>> source for libbpf so will this not be  considered as a new package
>> rather than a version change?
> 
> No. There is currently a libbpf0 binary package. After the source
> package split, there will still be a libbpf0 binary package. The
> version of that binary package cannot go backwards.
> 
> (The source package will indeed be NEW, but that's not the issue under
> discussion here.)

But, if I am correct, the source could be using a version without epoch
and only use the epoch in the binary package (which can be dropped if
libbpf0 is ever replaced by libbpf1).

Paul



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-05 Thread Adam D. Barratt
On Wed, 2020-02-05 at 22:42 +, Sudip Mukherjee wrote:
> On Wed, Feb 5, 2020 at 10:22 PM Christian Barcenas
>  wrote:
> > Because this changes the versioning scheme from kernel releases
> > (libbpf-dev and libbpf0 currently are at 5.4.13-1 in sid) to libbpf
> > version numbers (0.0.6-1), the epoch needs to be incremented to 1 I
> > believe.
> 
> I had this doubt but 5.4.13-1 is the linux source version, and
> libbpf0 has the version of 0.0.5. And since there is no separate
> source for libbpf so will this not be  considered as a new package
> rather than a version change?

No. There is currently a libbpf0 binary package. After the source
package split, there will still be a libbpf0 binary package. The
version of that binary package cannot go backwards.

(The source package will indeed be NEW, but that's not the issue under
discussion here.)

Regards,

Adam



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-05 Thread Sudip Mukherjee
On Wed, Feb 5, 2020 at 10:22 PM Christian Barcenas
 wrote:
>
> Because this changes the versioning scheme from kernel releases
> (libbpf-dev and libbpf0 currently are at 5.4.13-1 in sid) to libbpf
> version numbers (0.0.6-1), the epoch needs to be incremented to 1 I
> believe.

I had this doubt but 5.4.13-1 is the linux source version, and libbpf0 has the
version of 0.0.5. And since there is no separate source for libbpf so will this
not be  considered as a new package rather than a version change?


-- 
Regards
Sudip



Re: Bug#950760: RFS: libbpf/0.0.6-1 -- eBPF helper library (development files)

2020-02-05 Thread Christian Barcenas
Because this changes the versioning scheme from kernel releases
(libbpf-dev and libbpf0 currently are at 5.4.13-1 in sid) to libbpf
version numbers (0.0.6-1), the epoch needs to be incremented to 1 I
believe.

CC'ing debian-devel for discussion/consensus, per Debian Policy Manual 5.6.12.

Christian

On Wed, Feb 5, 2020 at 12:57 PM Sudip Mukherjee
 wrote:
>
> Package: sponsorship-requests
> Severity: wishlist
>
> Dear mentors,
>
> I am looking for a sponsor for my package "libbpf"
>
>  * Package name: libbpf
>Version : 0.0.6-1
>Upstream Author : NA
>  * URL : https://github.com/libbpf/libbpf
>  * License : LGPL-2.1
>  * Vcs : https://github.com/sudipm-mukherjee/libbpf
>Section : libs
>
> It builds those binary packages:
>
>   libbpf-dev - eBPF helper library (development files)
>   libbpf0 - eBPF helper library (shared library)
>
> To access further information about this package, please visit the following 
> URL:
>
>   https://mentors.debian.net/package/libbpf
>
> Alternatively, one can download the package with dget using this command:
>
>   dget -x 
> https://mentors.debian.net/debian/pool/main/libb/libbpf/libbpf_0.0.6-1.dsc
>
> Changes since the last upload:
>
>* Package from github. (Closes: #948041)
>
> Note:
> I think this should be done after 
> https://salsa.debian.org/kernel-team/linux/merge_requests/207
> has been merged.
>
>
> --
> Regards
> Sudip
>