Re: Fwd: calm: cygwin package upload report from sourceware.org for Chris Sutcliffe

[Jon Turney <jon dot turney at dronecode dot org dot uk>]

On 18/11/2016 17:37, Chris Sutcliffe wrote:
> On 18 November 2016 at 12:18, Jon Turney wrote:
> > On 18/11/2016 16:47, Chris Sutcliffe wrote:
> > > I've uploaded a new mksh build using the recently released cygport (v
> > > 0.23) and it seems like something isn't playing nice with the "upload"
> > > command and calm:
> > [...]
> > > Did I miss something in terms of a needed update to the mksh.cygport file?
> >
> >
> > You should have received the previous mail which indicates the reason the
> > upload was rejected.
>
> There was a .stackdump file that snuck in the x86_64 mksh release
> directory, which calm complained about, but I wasn't aware that it
> halted the upload process, and I was confused as to calm stopping the
> x86 release as well.

Sorry for the inconvenience.

After an error, it's very hard to be sure that allowing the upload to 
proceed won't lead to an undesired state, so I've chosen to be 
conservative and abandon the upload.

(On the specific point that an error in x86_64 shouldn't stop x86 being 
updated: this isn't completely true, for e.g. if a change to a noarch 
package results in a valid x86 packageset but an invalid x86_64 
packageset, we can't accept it, currently)

I'll look at improving the language in the upload report so it's clearer 
that nothing is going to be done after an error.

> I've since gone back and removed the .stackdump
> file and I've added a !ready file to both the x86_64 and x86 release
> directories, so hopefully things are correct now.

Yes, seems to have worked fine.