diff mbox series

[next,1/2] package/watchdogd: bump to version 3.5

Message ID 20211202013902.1593562-2-troglobit@gmail.com
State Accepted
Headers show
Series fix name clash between libite and LiTE | expand

Commit Message

Joachim Wiberg Dec. 2, 2021, 1:39 a.m. UTC
This addresses the autobuilder failure for master as per 2021-11-27.
The root cause was a name clash in a dependency, the libite library,
clashing with the DirectFB LiTE library header files.  Hence, this
update alone does not fix [1], libite also needs to be updated.  To
provide a smooth transition though, watchdogd should be updated first.

[1]: http://autobuild.buildroot.net/results/185c753af2aa159b494b13f78b0826dddbe4aed6

Signed-off-by: Joachim Wiberg <troglobit@gmail.com>
---
 package/watchdogd/watchdogd.hash | 2 +-
 package/watchdogd/watchdogd.mk   | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

Comments

Thomas Petazzoni Dec. 4, 2021, 8:04 p.m. UTC | #1
On Thu,  2 Dec 2021 02:39:01 +0100
Joachim Wiberg <troglobit@gmail.com> wrote:

> This addresses the autobuilder failure for master as per 2021-11-27.
> The root cause was a name clash in a dependency, the libite library,
> clashing with the DirectFB LiTE library header files.  Hence, this
> update alone does not fix [1], libite also needs to be updated.  To
> provide a smooth transition though, watchdogd should be updated first.
> 
> [1]: http://autobuild.buildroot.net/results/185c753af2aa159b494b13f78b0826dddbe4aed6
> 
> Signed-off-by: Joachim Wiberg <troglobit@gmail.com>
> ---
>  package/watchdogd/watchdogd.hash | 2 +-
>  package/watchdogd/watchdogd.mk   | 2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)

Thanks, I've applied both to next, as indicated in the patch title.
However, in the commit log, you seem to indicate that it's fixing a
build issue in master. Could you confirm?

Thanks,

Thomas
Joachim Wiberg Dec. 4, 2021, 8:23 p.m. UTC | #2
On 12/4/21 9:04 PM, Thomas Petazzoni wrote:
> On Thu,  2 Dec 2021 02:39:01 +0100
> Joachim Wiberg <troglobit@gmail.com> wrote:
>> This addresses the autobuilder failure for master as per 2021-11-27.
>> The root cause was a name clash in a dependency, the libite library,
>> clashing with the DirectFB LiTE library header files.  
> Thanks, I've applied both to next, as indicated in the patch title.
> However, in the commit log, you seem to indicate that it's fixing a
> build issue in master. Could you confirm?

Very unfortunate name clash between include files in two packages.

http://autobuild.buildroot.net/results/185c753af2aa159b494b13f78b0826dddbe4aed6


However, I think it's highly unlikely they'll be enabled at the same
time. My recommendation is to go for next and let 2021.11 out the door.

Best regards
 /Joachim
Thomas Petazzoni Dec. 4, 2021, 8:49 p.m. UTC | #3
On Sat, 4 Dec 2021 21:23:23 +0100
Joachim Wiberg <troglobit@gmail.com> wrote:

> Very unfortunate name clash between include files in two packages.

Yes, I had followed this.

> http://autobuild.buildroot.net/results/185c753af2aa159b494b13f78b0826dddbe4aed6
> 
> However, I think it's highly unlikely they'll be enabled at the same
> time. My recommendation is to go for next and let 2021.11 out the door.

But if the build issue exists in master, we need to fix it in master,
or the issue will keep popping up again and again in the autobuilders.

Thomas
Yann E. MORIN Dec. 4, 2021, 8:59 p.m. UTC | #4
Thomas, All,

On 2021-12-04 21:23 +0100, Joachim Wiberg spake thusly:
> On 12/4/21 9:04 PM, Thomas Petazzoni wrote:
> > On Thu,  2 Dec 2021 02:39:01 +0100
> > Joachim Wiberg <troglobit@gmail.com> wrote:
> >> This addresses the autobuilder failure for master as per 2021-11-27.
> >> The root cause was a name clash in a dependency, the libite library,
> >> clashing with the DirectFB LiTE library header files.  
> > Thanks, I've applied both to next, as indicated in the patch title.
> > However, in the commit log, you seem to indicate that it's fixing a
> > build issue in master. Could you confirm?

We've discussed this on IRC with Joachim, and it was my expectation that
the patches would be aimed at master, because they *do* fix an actual
build failure in master.

The reasoning for accepting the bump in master was that libite is only
ever used by watchdogd in Buldroot. Both packages are maintained
upstream and in Buildroot by Joachim, which is pretty fast at addressing
build issues.

> Very unfortunate name clash between include files in two packages.
> http://autobuild.buildroot.net/results/185c753af2aa159b494b13f78b0826dddbe4aed6
> 
> However, I think it's highly unlikely they'll be enabled at the same
> time. My recommendation is to go for next and let 2021.11 out the door.

Even if that were true, it will still happen in our CI and thus cause
build failures, which creates noise and may hide other, more intersting
failures. Besides, once a package is in the tree, people are going to
use it.

DirectFB is often used in embedded, and watchdogd is a process
supervisor for embedded, and so it is not too far-fetched to see that
the two could be easily used on the same system. For real. We want to
avoid name-clashing for those.

So I would say that we should also apply them to master.

Regards,
Yann E. MORIN.
Joachim Wiberg Dec. 5, 2021, 7:53 a.m. UTC | #5
Hi everyone!

On 12/4/21 9:59 PM, Yann E. MORIN wrote:
> On 2021-12-04 21:23 +0100, Joachim Wiberg spake thusly:
>> On 12/4/21 9:04 PM, Thomas Petazzoni wrote:
>>> On Thu,  2 Dec 2021 02:39:01 +0100
>>> Joachim Wiberg <troglobit@gmail.com> wrote:
>>>> This addresses the autobuilder failure for master as per 2021-11-27.
>>>> The root cause was a name clash in a dependency, the libite library,
>>>> clashing with the DirectFB LiTE library header files.  
>>> Thanks, I've applied both to next, as indicated in the patch title.
>>> However, in the commit log, you seem to indicate that it's fixing a
>>> build issue in master. Could you confirm?
> We've discussed this on IRC with Joachim, and it was my expectation that
> the patches would be aimed at master, because they *do* fix an actual
> build failure in master.
> [snip]
> So I would say that we should also apply them to master.

Also, on 12/4/21 9:49 PM, Thomas Petazzoni wrote:
> On Sat, 4 Dec 2021 Joachim Wiberg <troglobit@gmail.com> wrote:
>> However, I think it's highly unlikely they'll be enabled at the same
>> time. My recommendation is for next and let 2021.11 out the door.
> But if the build issue exists in master, we need to fix it in master,
> or the issue will keep popping up again and again in the autobuilders.

I'm sorry for the misunderstanding on my part.  I simply didn't realize
how seriously you viewed the issue, and also did not want to needlessly
block the release.

I leave release critical decisions up to your discretion, obviously.

Best regards
 /Joachim
diff mbox series

Patch

diff --git a/package/watchdogd/watchdogd.hash b/package/watchdogd/watchdogd.hash
index 7c1acf0820..09a34d6b47 100644
--- a/package/watchdogd/watchdogd.hash
+++ b/package/watchdogd/watchdogd.hash
@@ -1,5 +1,5 @@ 
 # Upstream .sha256 from GitHub
-sha256 f36fdea9b323534a704c4c23d9af8050480bca4bdad89cdb50d26fcac316d172  watchdogd-3.4.tar.gz
+sha256 33ec4edc8cb4ada7a4d8324a27d897d33aae5e83257ed64d3d37508825c11237  watchdogd-3.5.tar.gz
 
 # Locally calculated
 sha256 fd685e20931174308c45a26418a7ce34d66704c4e4b92ab1d8299deb255cd676  LICENSE
diff --git a/package/watchdogd/watchdogd.mk b/package/watchdogd/watchdogd.mk
index 34b17aa169..b666b56433 100644
--- a/package/watchdogd/watchdogd.mk
+++ b/package/watchdogd/watchdogd.mk
@@ -4,7 +4,7 @@ 
 #
 ################################################################################
 
-WATCHDOGD_VERSION = 3.4
+WATCHDOGD_VERSION = 3.5
 WATCHDOGD_SITE = https://github.com/troglobit/watchdogd/releases/download/$(WATCHDOGD_VERSION)
 WATCHDOGD_LICENSE = ISC
 WATCHDOGD_LICENSE_FILES = LICENSE