port ${name}

[Ryan Schmidt]

On Oct 18, 2014, at 9:59 PM, Mark Brethen wrote:

> On Oct 18, 2014, at 9:50 PM, Ryan Schmidt wrote:
> 
>> On Oct 18, 2014, at 9:35 PM, Mark Brethen wrote:
>> 
>>> However an issue I've found is that if the install process is interrupted, a subsequent re-issue of 'port install reduce' will fail due to the fact that the subport that was interrupted isn't clean.
>> 
>> What error occurs?
> 
> I had canceled the install to make changes, and reissued the port install command. 
> 
> brethen-mbp:ports marbre$ sudo port install reduce
> Password:
> Portfile changed since last build; discarding previous state.
> --->  Computing dependencies for reduce
> --->  Dependencies to be installed: reduce-csl reduce-psl
> --->  Fetching archive for reduce-csl
> --->  Attempting to fetch reduce-csl-20101217_0.darwin_13.x86_64.tbz2 from http://packages.macports.org/reduce-csl
> --->  Attempting to fetch reduce-csl-20101217_0.darwin_13.x86_64.tbz2 from http://mse.uk.packages.macports.org/sites/packages.macports.org/reduce-csl
> --->  Attempting to fetch reduce-csl-20101217_0.darwin_13.x86_64.tbz2 from http://lil.fr.packages.macports.org/reduce-csl
> --->  Fetching distfiles for reduce-csl
> Error: org.macports.fetch for port reduce-csl returned: Subversion check out failed
> Error: Failed to install reduce-csl
> Please see the log file for port reduce-csl for details:
>    /opt/local/var/macports/logs/_Users_marbre_ports_math_reduce/reduce-csl/main.log
> Error: The following dependencies were not installed: reduce-csl reduce-psl
> To report a bug, follow the instructions in the guide:
>    http://guide.macports.org/#project.tickets
> Error: Processing of port reduce failed
> 
> At this point I issued 'sudo port clean reduce-csl' and everything was ok, but should portfiles anticipate such an event?

Ah, so you cancelled a Subversion fetch. Yes, it's normal for a second Subversion fetch on top of an incomplete one to fail; don't interrupt the fetch phase, or if you do, clean before trying again. Ports are not expected to individually handle this situation. If you believe it's a common enough problem that it needs to be handled at all, we could modify base to remove existing directories before a Subversion fetch.

This issue may apply to other revision control systems too, I'm not sure.

There is work in progress by Larry to cache these revision control system checkouts, which may affect how a solution to this issue gets implemented.