Update Python and Flask

[methane]

• Update Python
• Use hkp:// to get gpg key for environment in firewall.
• Install tornado in installer.py
• Add Flask-PyPy

[seanaedmiston]

Hi,

I took a look at your pull request - I am very interested to see how PyPy works. I did however notice that in the 'Fortunes' test you have not added the additional Fortune at request time as specified (requirement #5 here: http://www.techempower.com/benchmarks/#section=code). Would be good if you could add this too.

Sean

[methane]

@seanaedmiston Is it fixed?

[seanaedmiston]

Looks like it will do the job.

One minor point. In my implementation of the 'raw' version, I found that it
performed noticeably better if instead of creating a new Fortune object
for each row, the row data was loaded into a list of tuples. (If you are
not going to use an ORM it didn't seem to make a lot of sense to create a
bunch of new objects for this simple use case.) The downside of this is
that the elements cannot be referenced by name (ie. fortune.id,
fortune.message) and so a different template is needed which references by
element (i.e. fortune[0], fortune[1]).

Cheers

On Mon, May 13, 2013 at 5:44 PM, INADA Naoki [email protected]:

@seanaedmiston https://github.com/seanaedmiston Is it fixed?

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/269#issuecomment-17796770
.

[methane]

@seanaedmiston I've misunderstand what "Within the scope of the request, a new Fortune object must be constructed and added to the list." means. I've rollbacked 77dda2a.

[bhauer]

The intent of the addition of a new Fortune object during the scope of each request is:

1. To demonstrate a dynamically-sized list/array data structure. The size needs to grow by one to make room for a newly constructed Fortune object.
2. Demonstrate the construction of a new object.
3. To enforce that sorting is done within the application code rather than within a ORDER BY clause in the query.

Because of number 2 above, we strongly prefer the use of a Fortune class and instance objects even if you are not using an ORM. To my eye, using an array or tuple structure seems very aggressive optimization, making it atypical for a production application. Am I wrong there?

[pfalls1]

@methane The fortunes test is failing because there's no template fortunes.html

[methane]

@pfalls1 Sorry, I've added it now.

[methane]

@bhauer dbraw_engine is wrapper of raw MySQLdb (or PyMySQL) driver.
It receives tuples from underlaying driver and wraps it with new SQLAlchemy's row object.
This wrapped object provides .message property and template use it.
So I think dbraw_engine meets "creating new object" requirements.

[pfalls1]

@methane Looks great, thanks!