tapestry-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Howard M. Lewis Ship (JIRA)" <j...@apache.org>
Subject [jira] Commented: (TAP5-577) TAP5-422 changes break persistent locale backwards compatibility.
Date Thu, 12 Mar 2009 18:22:50 GMT

    [ https://issues.apache.org/jira/browse/TAP5-577?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12681430#action_12681430

Howard M. Lewis Ship commented on TAP5-577:

I'm not sure if this really qualifies as breaking backwards compatibility, since so much of
this behavior was internal in Tapestry 5.0.

The key features haven't changed:

- A users locale is initially determined from browser headers
- A specific locale may be selected via PersistentLocale.set()
- After invoking set(), later requests ignore the header and use the preferred locale

The big difference is that 5.0 used a client-side cookie, and 5.1 (by default) encodes the
locale into the URL path.

Assuming users don't care about URLs, they won't see any difference ... except that when first
accessing the 5.1 application they may revert to the default locale (determined from browser
headers). However, this might occur anyway, if their cookie was deleted or expired.

The remaining difference is in how the locale is persisted between requests. Tapestry 5.1
already includes the ability to turn off encoding of the locale into the URL path. 

You three steps should work in theory.

What we have here is a change in behavior but not functionality; i.e., a change in the implementation
of persistent locales. I don't see this as a backwards compatibility problem as the vast majority
of users (who were not mucking about in the internals of Tapestry to customize persistent
locales) will not see a difference in the upgrade (just slightly different URLs).

> TAP5-422 changes break persistent locale backwards compatibility.
> -----------------------------------------------------------------
>                 Key: TAP5-577
>                 URL: https://issues.apache.org/jira/browse/TAP5-577
>             Project: Tapestry 5
>          Issue Type: Bug
>          Components: tapestry-core
>    Affects Versions:,
>            Reporter: Andy Blower
>            Priority: Critical
> I think that the changes made in T5.1 for TAP5-422 break backwards compatibility with
T5.0's locale persistence. In T5.0 it was a simple matter to override the default cookie persistence
by creating a custom implementation of the PersistentLocale service and contributing it to
be used instead of the standard internal T5 implementation.
> The TAP5-422 changes broke backwards compatibility because anyone who's created their
own implementation of PersistentLocale, or just wants the 5.0 cookie persistence behaviour,
would have found that it's a lot more work and involves some heavy changes to Tapestry internals.
Now with the recent changes for TAP5-418 (committed yesterday), the situation had been alleviated
somewhat by allowing the the hard-wired URl locale persistence to be switched off using a
new symbol.
> However, I still think that this breaks backwards compatibility in two ways:
> 1) By changing the default behaviour of locale persistence so that anyone relying on
the locale persistence behaviour of 5.0 will have to make non-trivial changes when they upgrade
to 5.1 to keep the same operation.
> 2) By requiring so much work for anyone wanting to keep the 5.0 cookie persistence behaviour
or define their own custom locale persistence. (In 5.0 it was easy to figure out and implement
a custom locale persistence method)
> From my analysis of the changes made by TAP5-422 & TAP5-418, I think anyone wanting
non-URL based locale persistence will need to do the following when upgrading from 5.0 to
> 1) Set the ENCODE_LOCALE_INTO_PATH symbol to false.
> 2) Create an implementation of PersistentLocale and contribute it to the IOC. (copied
from the standard 5.0 code if the old default cookie persistence is desired)
> 3) Create a custom filter written and created to do the same job as the 5.0 LocalizationFilter
and contribute it to the IOC RequestHandler. This filter will need to call the LocalizationSetter
setLocaleFromLocaleName() method instead of the old setThreadLocale() method. 
> My suggested resolution would be to re-instate the 5.0 cookie persistence (LocalizationFilter
& PersistentLocaleImpl) and have the new ENCODE_LOCALE_INTO_PATH symbol default to false
allowing 5.1 to work the same way as 5.0 out of the box. If the symbol is set to true, then
the LocalizationFilter is disabled (not contributed to RequestHandler) and the PersistentLocale
service will need to just store the locale (not set it in a cookie) for later use by LinkSourceImpl.
LocalizationSetterImpl.setLocaleFromLocaleName(String localeName) would also need changing
back to overriding the passed localeName if a persistent one had been set into the PersistentLocale
service. There may by a much better solution than this as I've not spent much time on it,
but I though I should try to be helpful as possible.
> (It should be noted that this is purely a product of my analysis of the 5.1 code, I have
not found the time to actually run T5.1 and test this out - I should be able to do this in
about a week and a half, but I'm currently approaching a major milestone deadline. Hopefully
someone else will find the time to prove or disprove my hypothesis.)

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message