pause icon indicating copy to clipboard operation
pause copied to clipboard

Email sent to author should refer to fast matrix

Open khwilliamson opened this issue 6 years ago • 6 comments

The email I just received: CPAN Testers will start reporting results in an hour or so:

http://matrix.cpantesters.org/?dist=Pod-Simple

should instead have had the link

http://fast-matrix.cpantesters.org/?dist=Pod-Simple

khwilliamson avatar Oct 26 '19 17:10 khwilliamson

In the ticket as displayed here, at https://github.com/andk/pause/issues/345 , I cannot not discover any reasoning. Maybe I'm missing something. Please elaborate.

andk avatar Oct 27 '19 07:10 andk

I waited and waited for results to show up on the sent link (more than the hour or two promised). Nothing did. On irc, people told me about the second link, and sure enough it had results.

The second link is superior in terms of getting fast feedback about an upload, which is what an author wants. Why not inform them of the superior link?

khwilliamson avatar Oct 27 '19 12:10 khwilliamson

There's more to say about the relation between matrix and fast-matrix:

  • both try to be correct and fast
  • when they both succeed, there is hardly a difference
  • the matrix is the canonical representation
  • the fast matrix often is a stop gap solution when the matrix has a bug or outage
  • when the fast matrix falls out of sync, it has to wait for the matrix to provide the truth
  • there are already links from every page of the matrix to the corresponding page of the fast matrix and vice versa

Weighing all these facts together leads me to think that a link to the matrix is the best choice. When other facts can be provided, we may reconsider what we want to change.

andk avatar Oct 28 '19 19:10 andk

Can we give them both links, and explain that the fast-matrix will give faster results initially, but no links to the reports, and the more complete (but slower) results are available at the regular link?

karenetheridge avatar Oct 28 '19 20:10 karenetheridge

Glad you ask! Caveats:

  • it's not true that the fast matrix has no links to the reports,
  • it's not true that the fast matrix is always faster, not even initially.

andk avatar Oct 29 '19 06:10 andk

when the fast matrix falls out of sync, it has to wait for the matrix to provide the truth

It's even worse: if the fast-matrix scripts fail to parse bits of tail.txt, then it never recovers, there will always be a gap.

Actually I think an additional link to fast-matrix would be OK, if it's accompanied by some explaining text.

eserte avatar Jul 04 '20 07:07 eserte