Bug#790103: new default build flag from dpkg: -Wdate-time

2015-11-19 Thread Georges Khaznadar
Hello,

Mattia Rizzolo a écrit :
[...]
> > > > That means avr-gcc will be upgraded to the atmel release 3.5.0 based on
> > > gcc
> > > > 4.9.2 soon.
> > > soon = ?
> > Within a week or three...
> 
> That's totally fine.  This thing won't happen before for sure.

This is fine for me too!

Best regards,   Georges.



signature.asc
Description: PGP signature


Bug#790103: new default build flag from dpkg: -Wdate-time

2015-11-18 Thread Hakan Ardo
Hi,
the archlinux avr-gcc seams to be a standard unpatched gcc compiled for the
avr target. I would prefere to base the main avr-gcc package on the atmel
releases as those have historically been of higiher quality for avr usage.
That means avr-gcc will be upgraded to the atmel release 3.5.0 based on gcc
4.9.2 soon. Will that help? Otherwise my suggestion would be that we
introduce a new package, say avr-gcc5, that we base on the gnu release
5.2.0 and keep avr-gcc following the atmel releases.

On Sat, Nov 14, 2015 at 10:49 AM, Mattia Rizzolo  wrote:

> reassign 790103 gcc-avr 1:4.8.1+Atmel3.4.5-1
> retitle 790103 gcc-avr: base on a newer version of gcc
> affects 790103 expeyes
>
> On Sat, Nov 14, 2015 at 10:35:46AM +0100, Georges Khaznadar wrote:
> > > > 3- rebase the package gcc-avr on a newer version of gcc, which may
> be a
> > > >hard work. Such a work seems to exist, for example at
> > > >https://www.archlinux.org/packages/community/x86_64/avr-gcc/
> since
> > > >last July
> > >
> > > This is what should be done, really.
> >
> > I had a look at this last option, but I miss knowledge to keep on with
> > it: as I could guess, the current gcc-avr package is based on a frozen
> > archive of gcc-4.8, which is modified by adding subtle modifications.
> > Obviously, I can freeze an archive of gcc-5, but I cannot craft the
> > modifications to turn it into an efficient compiler for avr.
>
> I miss those too, fwiw.
>
> > So I shall reassign bug #790103 to gcc-avr, and keep the suggested link
> > to expeyes.
>
> done.
>
> > As a matter of fact, I cannot push the severity such a bugreport higher
> > than whishlist, as I cannot provide any help about the work to be done.
>
> This is not what is used to decide the severities :), but still it's a
> whishlist bug.
>
> If this is not fixed in time for the change I'll open another bug
> against expeyes to at least disable the flag.
> If for some reason (=> allow us to test your package, maybe?) you want
> to do it now adding this to d/rules should be sufficient
> DEB_BUILD_MAINT_OPTIONS=reproducible=-timeless
>
> > Best regards, Georges.
>
> enjoy!
>
> --
> regards,
> Mattia Rizzolo
>
> GPG Key: 66AE 2B4A FCCF 3F52 DA18  4D18 4B04 3FCD B944 4540  .''`.
> more about me:  http://mapreri.org  : :'  :
> Launchpad user: https://launchpad.net/~mapreri  `. `'`
> Debian QA page: https://qa.debian.org/developer.php?login=mattia  `-
>



-- 
Håkan Ardö


Bug#790103: new default build flag from dpkg: -Wdate-time

2015-11-18 Thread Mattia Rizzolo
On Wed, Nov 18, 2015 at 08:26:24PM +0100, Hakan Ardo wrote:
> That means avr-gcc will be upgraded to the atmel release 3.5.0 based on gcc
> 4.9.2 soon.

soon = ?

> Will that help? Otherwise my suggestion would be that we
> introduce a new package, say avr-gcc5, that we base on the gnu release
> 5.2.0 and keep avr-gcc following the atmel releases.

sounds a lot of work for very little gain.
Only one package would FTBFS with this, and there are other workarounds
as well.

-- 
regards,
Mattia Rizzolo

GPG Key: 66AE 2B4A FCCF 3F52 DA18  4D18 4B04 3FCD B944 4540  .''`.
more about me:  http://mapreri.org  : :'  :
Launchpad user: https://launchpad.net/~mapreri  `. `'`
Debian QA page: https://qa.debian.org/developer.php?login=mattia  `-


signature.asc
Description: PGP signature


Bug#790103: new default build flag from dpkg: -Wdate-time

2015-11-18 Thread Mattia Rizzolo
On Wed, Nov 18, 2015 at 09:40:51PM +0100, Hakan Ardo wrote:
> On Wed, Nov 18, 2015 at 8:31 PM, Mattia Rizzolo  wrote:
> > On Wed, Nov 18, 2015 at 08:26:24PM +0100, Hakan Ardo wrote:
> > > That means avr-gcc will be upgraded to the atmel release 3.5.0 based on
> > gcc
> > > 4.9.2 soon.
> > soon = ?
> Within a week or three...

That's totally fine.  This thing won't happen before for sure.

-- 
regards,
Mattia Rizzolo

GPG Key: 66AE 2B4A FCCF 3F52 DA18  4D18 4B04 3FCD B944 4540  .''`.
more about me:  http://mapreri.org  : :'  :
Launchpad user: https://launchpad.net/~mapreri  `. `'`
Debian QA page: https://qa.debian.org/developer.php?login=mattia  `-


signature.asc
Description: PGP signature


Bug#790103: new default build flag from dpkg: -Wdate-time

2015-11-18 Thread Hakan Ardo
On Wed, Nov 18, 2015 at 8:31 PM, Mattia Rizzolo  wrote:

> On Wed, Nov 18, 2015 at 08:26:24PM +0100, Hakan Ardo wrote:
> > That means avr-gcc will be upgraded to the atmel release 3.5.0 based on
> gcc
> > 4.9.2 soon.
>
> soon = ?
>

Within a week or three...


Bug#790103: new default build flag from dpkg: -Wdate-time

2015-11-14 Thread Mattia Rizzolo
reassign 790103 gcc-avr 1:4.8.1+Atmel3.4.5-1
retitle 790103 gcc-avr: base on a newer version of gcc
affects 790103 expeyes

On Sat, Nov 14, 2015 at 10:35:46AM +0100, Georges Khaznadar wrote:
> > > 3- rebase the package gcc-avr on a newer version of gcc, which may be a
> > >hard work. Such a work seems to exist, for example at
> > >https://www.archlinux.org/packages/community/x86_64/avr-gcc/ since
> > >last July
> > 
> > This is what should be done, really.
> 
> I had a look at this last option, but I miss knowledge to keep on with
> it: as I could guess, the current gcc-avr package is based on a frozen
> archive of gcc-4.8, which is modified by adding subtle modifications.
> Obviously, I can freeze an archive of gcc-5, but I cannot craft the
> modifications to turn it into an efficient compiler for avr.

I miss those too, fwiw.

> So I shall reassign bug #790103 to gcc-avr, and keep the suggested link
> to expeyes.

done.

> As a matter of fact, I cannot push the severity such a bugreport higher
> than whishlist, as I cannot provide any help about the work to be done.

This is not what is used to decide the severities :), but still it's a
whishlist bug.

If this is not fixed in time for the change I'll open another bug
against expeyes to at least disable the flag.
If for some reason (=> allow us to test your package, maybe?) you want
to do it now adding this to d/rules should be sufficient
DEB_BUILD_MAINT_OPTIONS=reproducible=-timeless

> Best regards, Georges.

enjoy!

-- 
regards,
Mattia Rizzolo

GPG Key: 66AE 2B4A FCCF 3F52 DA18  4D18 4B04 3FCD B944 4540  .''`.
more about me:  http://mapreri.org  : :'  :
Launchpad user: https://launchpad.net/~mapreri  `. `'`
Debian QA page: https://qa.debian.org/developer.php?login=mattia  `-


signature.asc
Description: PGP signature


Bug#790103: new default build flag from dpkg: -Wdate-time

2015-11-14 Thread Georges Khaznadar
Hi Mattia,

Mattia Rizzolo a écrit :
> On Thu, Nov 12, 2015 at 06:46:37PM +0100, Georges Khaznadar wrote:
> > Mattia Rizzolo a écrit :
> > > From our builds, we can only see 4 packages which fail to build due to
> > > this: [...]
> > > expeyes
> > > https://bugs.debian.org/790103
> [...] 
> > 3- rebase the package gcc-avr on a newer version of gcc, which may be a
> >hard work. Such a work seems to exist, for example at
> >https://www.archlinux.org/packages/community/x86_64/avr-gcc/ since
> >last July
> 
> This is what should be done, really.

I had a look at this last option, but I miss knowledge to keep on with
it: as I could guess, the current gcc-avr package is based on a frozen
archive of gcc-4.8, which is modified by adding subtle modifications.
Obviously, I can freeze an archive of gcc-5, but I cannot craft the
modifications to turn it into an efficient compiler for avr.

> > [...] The last [option] sounds like a good fix
> > rather than a mere workaround.
> 
> Maybe this is something (also) upstream should look at.
> 
> > Which is your opinion about this?
> 
> If everybody agrees on having avr-gcc deal with the whole CPPFLAGS is
> what it's supposed to do, then this bug should be reassigned to it, and
> a "affects" link be left.
> If this flag will be added before avr-gcc gets fixed, simply disabling
> reproducible/timeless is the best way forward.
> IMHO.

So I shall reassign bug #790103 to gcc-avr, and keep the suggested link
to expeyes.
As a matter of fact, I cannot push the severity such a bugreport higher
than whishlist, as I cannot provide any help about the work to be done.

Best regards,   Georges.

> 
> 
> -- 
> regards,
> Mattia Rizzolo
> 
> GPG Key: 66AE 2B4A FCCF 3F52 DA18  4D18 4B04 3FCD B944 4540  .''`.
> more about me:  http://mapreri.org  : :'  :
> Launchpad user: https://launchpad.net/~mapreri  `. `'`
> Debian QA page: https://qa.debian.org/developer.php?login=mattia  `-



-- 
Georges KHAZNADAR et Jocelyne FOURNIER
22 rue des mouettes, 59240 Dunkerque France.
Téléphone +33 (0)3 28 29 17 70



signature.asc
Description: PGP signature


Bug#790103: new default build flag from dpkg: -Wdate-time

2015-11-12 Thread Georges Khaznadar
Hi Mattia,

Mattia Rizzolo a écrit :
> ...
> We now think this is the right time to enable this flag by default.
> From our builds, we can only see 4 packages which fail to build due to
> this: [...]
> expeyes
> https://bugs.debian.org/790103

The reason why expeyes FTBFS, is that the flag -Wdate-time is applied to
a call of avr-gcc, which is based on gcc-4.8, and ignores this flag.

So I can imagine a few workarounds:
1- some trick to prevent the use of -Wdate-time when it touches avr-gcc
2- isolate the patch which introduced the new flag into gcc-4.8 and apply
   it to make a new release of the package gcc-avr_1:4.8.1+Atmel3
3- rebase the package gcc-avr on a newer version of gcc, which may be a
   hard work. Such a work seems to exist, for example at
   https://www.archlinux.org/packages/community/x86_64/avr-gcc/ since
   last July

I may try the first workaround, just replacing manually $(CFLAGS) by
a hardwired option chain in the Makefile which calls avr-gcc.

The second and the third workarounds should imply Hakan Ardo, who
maintains the package gcc-avr. The last one sounds like a good fix
rather than a mere workaround.

Which is your opinion about this?

Best regards,   Georges.



signature.asc
Description: PGP signature


Bug#790103: new default build flag from dpkg: -Wdate-time

2015-11-12 Thread Mattia Rizzolo
On Thu, Nov 12, 2015 at 06:46:37PM +0100, Georges Khaznadar wrote:
> Mattia Rizzolo a écrit :
> > From our builds, we can only see 4 packages which fail to build due to
> > this: [...]
> > expeyes
> > https://bugs.debian.org/790103
> 
> The reason why expeyes FTBFS, is that the flag -Wdate-time is applied to
> a call of avr-gcc, which is based on gcc-4.8, and ignores this flag.

ok, this would also be interesting to be forwarded on that bug, though.
I bounced the mail to the bug and added it to CC here

> So I can imagine a few workarounds:
> 1- some trick to prevent the use of -Wdate-time when it touches avr-gcc

no, please, no need to be so dirt :)

> 2- isolate the patch which introduced the new flag into gcc-4.8 and apply
>it to make a new release of the package gcc-avr_1:4.8.1+Atmel3

sounds wrong...

> 3- rebase the package gcc-avr on a newer version of gcc, which may be a
>hard work. Such a work seems to exist, for example at
>https://www.archlinux.org/packages/community/x86_64/avr-gcc/ since
>last July

This is what should be done, really.

> I may try the first workaround, just replacing manually $(CFLAGS) by
> a hardwired option chain in the Makefile which calls avr-gcc.

No need, remember that the default compiling flags are entirely
overridable at build time, see dpkg-buildflags(1).
This is anyway not really needed *right now* (unless you'd like to see
your package built in our CI), it'll be really needed once dpkg
incorporate it.

> The second and the third workarounds should imply Hakan Ardo, who
> maintains the package gcc-avr. The last one sounds like a good fix
> rather than a mere workaround.

Maybe this is something (also) upstream should look at.

> Which is your opinion about this?

If everybody agrees on having avr-gcc deal with the whole CPPFLAGS is
what it's supposed to do, then this bug should be reassigned to it, and
a "affects" link be left.
If this flag will be added before avr-gcc gets fixed, simply disabling
reproducible/timeless is the best way forward.
IMHO.


-- 
regards,
Mattia Rizzolo

GPG Key: 66AE 2B4A FCCF 3F52 DA18  4D18 4B04 3FCD B944 4540  .''`.
more about me:  http://mapreri.org  : :'  :
Launchpad user: https://launchpad.net/~mapreri  `. `'`
Debian QA page: https://qa.debian.org/developer.php?login=mattia  `-


signature.asc
Description: PGP signature