Re: [BUG] zsystem:34: flock: invalid timeout value: '0'

[Daniel Shahaf <d.s@xxxxxxxxxxxxxxxxxx>]

Sebastian Gniazdowski wrote on Fri, 26 Jun 2020 21:42 +0200:
> On Fri, 26 Jun 2020 at 16:16, Daniel Shahaf <d.s@xxxxxxxxxxxxxxxxxx> wrote:
> > Sebastian Gniazdowski wrote on Thu, 25 Jun 2020 19:49 +0200:  
> > > Hi
> > > I think that something happened to zsystem / flock, as the following
> > > returns the above error under Cygwin:
> > >
> > > $ zsystem flock -t 0 -f FD -e /home/SG/.zsh_nr1
> > > zsystem:34: flock: invalid timeout value: '0'
> > >
> > > Timeout of value 0 is an acceptable one - it means no waiting for the  
> > > lock,  
> > > just try-to-lock and exit.  
> >
> > You've neglected to identify the environment it works in  
> 
> 
> What do you mean? I don't have access to environment other than Cygwin

Your readers don't know this.

> and I've written about the Cygwin-environment to narrow this report.

Yes, but you didn't state the relevant details (version numbers of zsh,
Cygwin, and Windows) of either the working case or the non-working case,
nor the OS name of the non-working case.

> > and the reason you didn't do
> > a bisect as courtesy prescribes.
> >
> > In this case, system.c gets so few commits that a bisect is almost
> > a formality.  The bug is likely to have been introduced by workers/45708.
> >  
> 
> I saw the subsecond timeout changes conversation/patch and assumed that
> it's the cause, thus I've left the interested party to be alerted by the
> title.

It's not clear to me what made you assume that patch was the cause.

At any rate, what you should have done is triage your own bug report and
state your findings.  What you had in fact done is posted a bug report
and left it for others to triage.  The part in the middle where you say
"I have triaged this and I think this is caused by 45708" makes all the
difference.  If you don't say it, it won't have happened as far as your
readers are concerned.

> > I recommend that you test the behaviour with and without that commit
> > and reported the results.
> > 
> > I further recommend that, if you have time, you also prepare a patch,
> > or at least an expected-to-fail regression test (using the 'f' flag
> > after the exit code).
> >  
> 
> I might prepare a patch, but I also think that the involved party might do
> this.

+1

I'm not going to comment any further about the process aspects.
Regarding the actual bug report, the next steps are for the
aforementioned commit to be confirmed as causing an unintended
behaviour change and for a test and a patch to be written.

Daniel