You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@whimsical.apache.org by Shane Curcuru <as...@shanecurcuru.org> on 2019/05/31 11:14:47 UTC
Roster tool required data use?
Attempting to run the roster tool locally, after having *thought* I set
everything up for it, I get:
Unable to find svn checkout for https://svn.apache.org/repos/member_apps
From lib/whimsy/asf/memapps.rb:12
As a specific question and a general one: is there any way to make basic
roster functionality degrade gracefully, instead of just dying before it
displays even the welcome page?
I suppose it's only an issue for local developers, since the server
should always have all-the-repos. But it somehow feels like scripts for
non-critical data - like member application PDFs - should silently fail
unless someone's actually attempting to edit that data.
The larger question is: are there scripts that could degrade when an
ASF::SVN call fails, instead of just barfing? I.e. catch whatever
errors might happen, and then degrade performance when attempting to use
those resources later, instead of completely dying. I guess this is
just a general reminder to ponder how we do holistic error checking.
Anyway, this probably isn't important, and after checking out
member_apps, my whimsy.local/roster now dies with a 502 error, so
there's something else to fix instead! 8-)
--
- Shane
Whimsy PMC
The Apache Software Foundation
Re: Roster tool required data use?
Posted by Sam Ruby <ru...@intertwingly.net>.
On Fri, May 31, 2019 at 7:14 AM Shane Curcuru <as...@shanecurcuru.org> wrote:
>
> Attempting to run the roster tool locally, after having *thought* I set
> everything up for it, I get:
>
> Unable to find svn checkout for https://svn.apache.org/repos/member_apps
>
> From lib/whimsy/asf/memapps.rb:12
>
> As a specific question and a general one: is there any way to make basic
> roster functionality degrade gracefully, instead of just dying before it
> displays even the welcome page?
>
> I suppose it's only an issue for local developers, since the server
> should always have all-the-repos. But it somehow feels like scripts for
> non-critical data - like member application PDFs - should silently fail
> unless someone's actually attempting to edit that data.
>
> The larger question is: are there scripts that could degrade when an
> ASF::SVN call fails, instead of just barfing? I.e. catch whatever
> errors might happen, and then degrade performance when attempting to use
> those resources later, instead of completely dying. I guess this is
> just a general reminder to ponder how we do holistic error checking.
https://whimsy.apache.org/docs/api/ASF/SVN.html
ASF::SVN['x'] throws an exception if 'x' is not found.
ASF::SVN.find('x') returns nil if 'x' is not found.
ASF::SVN.find was there first, and apps either degraded gracefully or
blew up trying to dereference nil. Over time, I found degrading
gracefully in most cases was annoying as things would work differently
locally without explaining why. And an explicit error was better than
a nil pointer exception.
> Anyway, this probably isn't important, and after checking out
> member_apps, my whimsy.local/roster now dies with a 502 error, so
> there's something else to fix instead! 8-)
>
> --
>
> - Shane
> Whimsy PMC
> The Apache Software Foundation