It's been a few days since anyone has responded, so I went ahead and ran
the numbers. To recap, the question was to rank how much focus should
be given to stylistic PSR-2-esque specs (what I'll call a "soft spec")
vs. interfaces and PSR-3-esque specs (a "hard spec"). 1 was "all soft",
10 was "all hard" , and 5 was "equal weight".
The results are here:
https://docs.google.com/spreadsheet/ccc?key=0AsMrMKNHL1uGdHh1bDRnWVU3YjAyNWFmb0VGTHkzUGc&usp=sharing
The first sheet has the raw data, the second sheet some numbers I've
calculated off of it.
Data crunching of note:
- 18 of 27 voting members replied.
- 53 non-voting members replied.
- There were only *3* people total who responded with a score less than
5 (meaning equal weight). All three were non-voting respondents. Two
of them self-identified as being Zend-centric, one as being
Symfony-centric. None of them were voting members.
- The overall average rating was about 7.3. The median was 8. That was
fairly consistent between voting and non-voting members.
- The most common vote was 5 (16 votes), right down the middle. But it
was swamped by the number of 8s and 9s (12 each).
- Broken out by the project people identified with (which in some cases
was multiple projects):
-- Symfony was the most-represented project by a wide margin, although
it was beaten out by non-aligned.
-- CakePHP was the most "hard", with 10s across the board.
-- Zend was the most "soft", with an average of 4.2 from 5 votes. It was
the only project to average less than 5.
- The lowest per-project median was 5. Most were 8s and 9s.
(Special thanks to Michael Favia of Drupal for showing me how to do
anything useful in Google Spreadsheet formulas!)
Conclusions:
(This is the subjective interpretation part. The stuff above is just data.)
- The voting members and non-members seem to be pretty much in agreement
about where our focus should be, give or take some outliers.
- That focus is squarely on the "Focus on interfaces and similar 'hard'
specs, but yeah 'soft' specs are kinda useful, too."
- That may or may not have been the intent of the people who started
this group originally. I do not presume to speak for them. In the here
and now, though, the interest is decidedly in favor of common
cross-project interfaces and other run-time-impacting specs. Therefore,
that is where we should be putting our focus, and our marketing.
In short, this poll didn't show an ideal consensus but it did show a
very clear, strong preference for "hard" specs. Let's focus on that,
and keep that in mind when similar discussions come up in the future.
And with that, let's get back to the cache spec, the http spec, and
similar work that it seems clear is where we should be focusing.
Onward!
--Larry Garfield