Add string versions of User/Pickled objects to datastore models

Add string versions of User/Pickled objects to datastore models

On newly created objects Package.uploaders,
PackageVersion.uploader/readme/changelog will have corresponding String
variants.

After a batch processing update of old models, we can enforce the usage of the
new properties (and drop the old ones).

R=sgjesse@google.com

Committed: https://github.com/dart-lang/pub-dartlang/commit/16efcbb5a41e916d0aea3cbe109faf2224bf1a11

[kustermann, 2014-12-18 14:28:17]

kustermann@google.com changed reviewers:
+ sgjesse@google.com

[kustermann, 2014-12-18 14:28:17]

I plan on uploading a new appengine version and use this version as
PUB_HOSTED_URL and publish a dummy package.

If that works, I'll do batch processing of old data and make it the default
version.


As a second step, we'll change the code to use the new properties. 

As a third step, we'll remove the old properties completely.

[kustermann, 2014-12-18 14:28:24]

kustermann@google.com changed reviewers:
+ wibling@google.com

[Søren Gjesse, 2014-12-18 14:47:37]

lgtm

https://chromiumcodereview.appspot.com/816693002/diff/1/app/models/package_ve...
File app/models/package_version.py (right):

https://chromiumcodereview.appspot.com/816693002/diff/1/app/models/package_ve...
app/models/package_version.py:91: if self.readme:
Not related to this change, but in general we don't in any way control the
encoding of the README and CHANGELOG files in the tar. I guess we are hoping for
UTF-8?

[kustermann, 2014-12-18 15:02:57]

Patchset #2 (id:20001) has been deleted

[kustermann, 2014-12-18 15:04:41]

Please take another look if you want to, otherwise I'll start landing this.

https://chromiumcodereview.appspot.com/816693002/diff/1/app/models/package_ve...
File app/models/package_version.py (right):

https://chromiumcodereview.appspot.com/816693002/diff/1/app/models/package_ve...
app/models/package_version.py:91: if self.readme:
On 2014/12/18 14:47:37, Søren Gjesse wrote:
> Not related to this change, but in general we don't in any way control the
> encoding of the README and CHANGELOG files in the tar. I guess we are hoping
for
> UTF-8?

We're not just hoping, we're assuming:
 41         filename = min(readmes, key=lambda(n): (n.count('.'), n))
 42         text = unicode(tar.extractfile(filename).read(),
 43                        encoding='utf-8', errors='replace')
 44         return Readme(text, filename)


I've changed the code now to write it definitly as utf-8.

[Søren Gjesse, 2014-12-18 15:14:36]

LGTM

[kustermann, 2014-12-18 15:20:02]

Committed patchset #2 (id:40001) manually as
16efcbb5a41e916d0aea3cbe109faf2224bf1a11 (presubmit successful).