Bug #55452 for CPAN: upgrade stops at the first failure

Wed Mar 10 17:13:53 2010 NKH [...] cpan.org - Ticket created

Subject:

upgrade stops at the first failure

Command 'upgrade' stops at the first failing module installation making it more or less totally useless.

Wed Mar 10 17:21:34 2010 NKH [...] cpan.org - Correspondence added

also broken in CPAN-1.94_56

Wed Mar 10 17:21:42 2010 NKH [...] cpan.org - Status changed from 'new' to 'open'

Wed Mar 10 18:11:17 2010 dagolden [...] cpan.org - Correspondence added

Subject:	Re: [rt.cpan.org #55452] upgrade stops at the first failure
Date:	Wed, 10 Mar 2010 18:10:29 -0500
To:	bug-CPAN <bug-CPAN [...] rt.cpan.org>
From:	David Golden <dagolden [...] cpan.org>

On Wed, Mar 10, 2010 at 5:13 PM, NKH via RT <bug-CPAN@rt.cpan.org> wrote: Show quoted text

> Command 'upgrade' stops at the first failing module installation making > it more or less totally useless.

What is the result of "o conf halt_on_failure" ? -- David

Thu Mar 11 02:35:54 2010 NKH [...] cpan.org - Correspondence added

Show quoted text

> What is the result of "o conf halt_on_failure" ?

The result is '1'. If setting it to '0' would fix this problem, I wonder if 'upgrade' shouldn't have a separate configuration variable.

Thu Mar 11 08:22:39 2010 dagolden [...] cpan.org - Correspondence added

Subject:	Re: [rt.cpan.org #55452] upgrade stops at the first failure
Date:	Thu, 11 Mar 2010 06:34:35 -0500
To:	bug-CPAN [...] rt.cpan.org
From:	David Golden <dagolden [...] cpan.org>

On Thu, Mar 11, 2010 at 2:35 AM, NKH via RT <bug-CPAN@rt.cpan.org> wrote: Show quoted text

> Queue: CPAN > Ticket <URL: https://rt.cpan.org/Ticket/Display.html?id=55452 > >

>> What is the result of "o conf halt_on_failure" ?

> > The result is '1'. > > If setting it to '0' would fix this problem, I wonder if 'upgrade' > shouldn't have a separate configuration variable.

I think we have enough configuration variables already. Andreas -- my thoughts: (1) Just locally disable it (potential surprise if someone wants the behavior) (2) Warn the user (but possibly lost in scrollback) (3) Prompt -- "You've request halt_on_failure -- do you want to disable it during 'upgrade'"? -- David

Thu Mar 11 18:21:15 2010 NKH [...] cpan.org - Correspondence added

Show quoted text

> I think we have enough configuration variables already.

I agree. Show quoted text

> Andreas -- my thoughts: > > (1) Just locally disable it (potential surprise if someone wants the > behavior)

I firmly believe in the principle of least surprise so I don't like this solution at all. Show quoted text

> (2) Warn the user (but possibly lost in scrollback)

This is also a surprise. No? Show quoted text

> (3) Prompt -- "You've request halt_on_failure -- do you want to > disable it during 'upgrade'"?

This one gets my vote. Maybe a question that doesn't expect the user to know what 'halt_on_failure' means would be better. IE: it was set by default for me and you told me about it. I doubt all cpan user will get the same exceptional 'level of service'. I also believe that an upgrade status should be given after the upgrade as well as a link to a log file containing all the output generated during the upgrade. Btw, does cpan log anything during its normal (non upgrade) usage? Cheers, Nadim.

Thu Mar 11 18:36:28 2010 dagolden [...] cpan.org - Correspondence added

Subject:	Re: [rt.cpan.org #55452] upgrade stops at the first failure
Date:	Thu, 11 Mar 2010 18:35:31 -0500
To:	bug-CPAN <bug-CPAN [...] rt.cpan.org>
From:	David Golden <dagolden [...] cpan.org>

On Thu, Mar 11, 2010 at 6:21 PM, NKH via RT <bug-CPAN@rt.cpan.org> wrote: Show quoted text

> This one gets my vote. Maybe a question that doesn't expect the user to > know what 'halt_on_failure' means would be better. IE: it was set by > default for me and you told me about it. I doubt all cpan user will get > the same exceptional 'level of service'.

halt_on_failure defaults to 0. So you or a sysadmin must have set it to 1. David

Thu Mar 11 23:52:50 2010 ANDK [...] cpan.org - Correspondence added

I like none of the three options. What about (4) whenever halt_on_failure causes a halt it must indicate that in the fail summary. That way every user who has configured halt_on_failure once upon the time for some reason gets what he asked for but is also reminded about the configuration.

Fri Mar 12 00:01:11 2010 ANDK [...] cpan.org - Correspondence added

I mean, currently I see: Running make install make test had returned bad status, won't install without force Stopping: 'install' failed for 'M/MI/MIYAGAWA/Plack-0.9915.tar.gz'. Failed during this command: MIYAGAWA/Plack-0.9915.tar.gz : make_test NO and the "Stopping" line is red. It should probably read: Halting last command (configuration has halt_on_failure set): ...

Tue May 25 03:31:07 2010 NKH [...] cpan.org - Correspondence added

Any chance this will be fixed soon? I'd jump in if I knew where to start.

Tue May 25 07:41:04 2010 dagolden [...] cpan.org - Correspondence added

Subject:	Re: [rt.cpan.org #55452] upgrade stops at the first failure
Date:	Tue, 25 May 2010 07:40:34 -0400
To:	bug-CPAN [...] rt.cpan.org
From:	David Golden <dagolden [...] cpan.org>

On Tue, May 25, 2010 at 3:31 AM, NKH via RT <bug-CPAN@rt.cpan.org> wrote: Show quoted text

> Queue: CPAN > Ticket <URL: https://rt.cpan.org/Ticket/Display.html?id=55452 > > > Any chance this will be fixed soon? > > I'd jump in if I knew where to start.

Grep for 'halt_on_failure' in CPAN::Shell and send patch with a more sensible error message? -- David