| Message ID | 20240910073402.23281-1-ceggers@arri.de |
|---|---|
| State | Changes Requested |
| Headers | show |
| Series | [meta-swupdate] swupdate-progress: remove misleading 'exec' statement | expand |
Hi Christian, On 10.09.24 09:34, Christian Eggers wrote: > As this script fragment is sourced by another script (swupdate.sh), we > do not want to replace the current shell by swupdate-progress. The shell > indeed doesn't do this due to the '&' at the end of the command. So the > 'exec' statement seems to have no effect and should be removed in favor > of clarity. > Let's see, because this does not happen very often. 90-start-progress is not installed on systems with Systemd as virtual/init_manager. It remains as fallback with SystemV (or other init systems) and just in that case it is started by system.sh. But this is maybe the cause. Instead of this, does it makes more sense to have a swupdate-progress init script (with own start/stop/restart) ? swupdate.sh is thought to start just the SWUpdate daemon, and it starts the progress because in the past they were started from the same script, probably due to laziness (from me...). Having a single SystemV script running two daemons is pretty unusual and I can just imagine I did this because I was on hurry. Best regards, Stefano Babic > Signed-off-by: Christian Eggers <ceggers@arri.de> > --- > Please apply to all branches. > > recipes-support/swupdate/swupdate/90-start-progress | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/recipes-support/swupdate/swupdate/90-start-progress b/recipes-support/swupdate/swupdate/90-start-progress > index 3b0eb5e7d7c9..d574ce5dd660 100644 > --- a/recipes-support/swupdate/swupdate/90-start-progress > +++ b/recipes-support/swupdate/swupdate/90-start-progress > @@ -1 +1 @@ > -exec /usr/bin/swupdate-progress -w -r & > +/usr/bin/swupdate-progress -w -r &
Hi Stefano, On Tuesday, 10 September 2024, 11:28:01 CEST, Stefano Babic wrote: > Hi Christian, > > On 10.09.24 09:34, Christian Eggers wrote: > > As this script fragment is sourced by another script (swupdate.sh), we > > do not want to replace the current shell by swupdate-progress. The shell > > indeed doesn't do this due to the '&' at the end of the command. So the > > 'exec' statement seems to have no effect and should be removed in favor > > of clarity. > > > > Let's see, because this does not happen very often. > > 90-start-progress is not installed on systems with Systemd as > virtual/init_manager. It remains as fallback with SystemV (or other init > systems) and just in that case it is started by system.sh. But this is > maybe the cause. In my case it's busybox init which starts swupdate.sh ... > > Instead of this, does it makes more sense to have a swupdate-progress > init script (with own start/stop/restart) ? swupdate.sh is thought to > start just the SWUpdate daemon, and it starts the progress because in > the past they were started from the same script, probably due to > laziness (from me...). Having a single SystemV script running two > daemons is pretty unusual and I can just imagine I did this because I > was on hurry. Of course, a separate init script for swupdate-progress would also make sense. As a bonus, swupdate-progress can be started AFTER the swupdate daemon which would make more sense. As I have just removed the swupdate-progress package from my system, the result doesn't matter for me. I only created this patch to increase clarity for other users. regards, Christian > > Best regards, > Stefano Babic > > > Signed-off-by: Christian Eggers <ceggers@arri.de> > > --- > > Please apply to all branches. > > > > recipes-support/swupdate/swupdate/90-start-progress | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/recipes-support/swupdate/swupdate/90-start-progress b/recipes-support/swupdate/swupdate/90-start-progress > > index 3b0eb5e7d7c9..d574ce5dd660 100644 > > --- a/recipes-support/swupdate/swupdate/90-start-progress > > +++ b/recipes-support/swupdate/swupdate/90-start-progress > > @@ -1 +1 @@ > > -exec /usr/bin/swupdate-progress -w -r & > > +/usr/bin/swupdate-progress -w -r & > >
On 10.09.24 12:52, Christian Eggers wrote: > Hi Stefano, > > On Tuesday, 10 September 2024, 11:28:01 CEST, Stefano Babic wrote: >> Hi Christian, >> >> On 10.09.24 09:34, Christian Eggers wrote: >>> As this script fragment is sourced by another script (swupdate.sh), we >>> do not want to replace the current shell by swupdate-progress. The shell >>> indeed doesn't do this due to the '&' at the end of the command. So the >>> 'exec' statement seems to have no effect and should be removed in favor >>> of clarity. >>> >> >> Let's see, because this does not happen very often. >> >> 90-start-progress is not installed on systems with Systemd as >> virtual/init_manager. It remains as fallback with SystemV (or other init >> systems) and just in that case it is started by system.sh. But this is >> maybe the cause. > > In my case it's busybox init which starts swupdate.sh ... > >> >> Instead of this, does it makes more sense to have a swupdate-progress >> init script (with own start/stop/restart) ? swupdate.sh is thought to >> start just the SWUpdate daemon, and it starts the progress because in >> the past they were started from the same script, probably due to >> laziness (from me...). Having a single SystemV script running two >> daemons is pretty unusual and I can just imagine I did this because I >> was on hurry. > > Of course, a separate init script for swupdate-progress would also make sense. +1 > As a bonus, swupdate-progress can be started AFTER the swupdate daemon > which would make more sense. +1 > > As I have just removed the swupdate-progress package from my system, the > result doesn't matter for me. I only created this patch to increase clarity > for other users. ok - I will write somewhere (maybe in improvements) that this should be solved, so that this won't be forgotten. Best regards, Stefano > > regards, > Christian > >> >> Best regards, >> Stefano Babic >> >>> Signed-off-by: Christian Eggers <ceggers@arri.de> >>> --- >>> Please apply to all branches. >>> >>> recipes-support/swupdate/swupdate/90-start-progress | 2 +- >>> 1 file changed, 1 insertion(+), 1 deletion(-) >>> >>> diff --git a/recipes-support/swupdate/swupdate/90-start-progress b/recipes-support/swupdate/swupdate/90-start-progress >>> index 3b0eb5e7d7c9..d574ce5dd660 100644 >>> --- a/recipes-support/swupdate/swupdate/90-start-progress >>> +++ b/recipes-support/swupdate/swupdate/90-start-progress >>> @@ -1 +1 @@ >>> -exec /usr/bin/swupdate-progress -w -r & >>> +/usr/bin/swupdate-progress -w -r & >> >> > > > >
diff --git a/recipes-support/swupdate/swupdate/90-start-progress b/recipes-support/swupdate/swupdate/90-start-progress index 3b0eb5e7d7c9..d574ce5dd660 100644 --- a/recipes-support/swupdate/swupdate/90-start-progress +++ b/recipes-support/swupdate/swupdate/90-start-progress @@ -1 +1 @@ -exec /usr/bin/swupdate-progress -w -r & +/usr/bin/swupdate-progress -w -r &
As this script fragment is sourced by another script (swupdate.sh), we do not want to replace the current shell by swupdate-progress. The shell indeed doesn't do this due to the '&' at the end of the command. So the 'exec' statement seems to have no effect and should be removed in favor of clarity. Signed-off-by: Christian Eggers <ceggers@arri.de> --- Please apply to all branches. recipes-support/swupdate/swupdate/90-start-progress | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)