[racket-dev] more planet2 feedback

From: Jay McCarthy (jay.mccarthy at gmail.com)
Date: Thu Jan 10 16:29:41 EST 2013

On Thu, Jan 10, 2013 at 2:18 PM, Danny Yoo <dyoo at hashcollision.org> wrote:
> Hi Jay,
> Some more usability feedback:
> 1.  When updating checksums of my existing packages, I can't tell what
> gets updated, as the page for:
>     https://plt-etc.byu.edu:9004/manage
> can look the same before and after "Update checksums", so I don't know
> if I actually did anything, or if the server timed out during the
> request.  I know that there's a red start next to new package, but if
> it was already new to begin with, before "Update checksums", then
> there's no visible delta as far as the Manage page is concerned.

Red stars go next to packages that were recently updated. If you go to
an individual details page, you can view the last edit/update times.


> Also, if it's possible to mark the "Update checksums" URL so that it's
> sent as a POST rather than a GET, that would be great.
> 2.  The login page makes the automatic assumption that if you're
> logging in as someone who doesn't exist, that it must be a
> subscription request.  I think this is the wrong default, as it is
> unforgiving of typos.  e.g. I put in 'dyoo' instead of
> 'dyoo at hashcollision.org' for my email, and now the system is sending
> spurious email out there.
> Perhaps add one additional stage to the interaction, where it says
> something like "I don't know this user: send a subscription request?"
> or so?
> 3.  We talked privately about the error message that shows up when
> uploading a package with a malformed name.  e.g. I tried uploading
> ragg with the name "ragg: a Racket AST Generator Generator", and got a
> stack trace error message from the web server of the form:
> ---
> Exception
> The application raised an exception with the message:
> planet2: Illegal character in name; only alphanumerics, plus '-' and
> '_' allowed: "ragg: a Racket AST Generator Generator"
> Stack trace:
> edit-details at:
>   line 424, column 2, in file
> /home/plt-etc/local/galaxy/meta/planet2-index/official/main.rkt
> ---
> I argue that the error message is too "low-level" when presented by
> the web application because it's the result of a stack trace, rather
> than the result of constraints on the form elements on the submission
> page.  Though the error message does correctly mention that it's the
> 'name' that is at fault, still the mechanism for reporting the error
> is fairly indistinguishable from any other internal error in a
> web-servlet.

Jay McCarthy <jay at cs.byu.edu>
Assistant Professor / Brigham Young University

"The glory of God is Intelligence" - D&C 93

Posted on the dev mailing list.