incubator-esme-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ethan Jewett <esjew...@gmail.com>
Subject Re: Tests for Original API
Date Fri, 08 Jan 2010 17:43:43 GMT
That's fair enough. One interesting thing to note about APIs though,
is that in some sense any change to them is a bug. So even if we fix a
deeper bug within ESME, which causes the API to start behaving
"correctly", it still has significant potential to break clients that
have come to depend on the incorrect behavior.

This is not to say that we should not fix issues. In fact, often times
it makes very good sense to break clients. Often the clients should
have been designed more robustly, only used documented interfaces, or
maybe it is just worth it to break even good clients for some
improvements.

Just something to keep in mind.

Ethan

On Fri, Jan 8, 2010 at 11:33 AM, Richard Hirsch <hirsch.dick@gmail.com> wrote:
> My idea is that through testing the old API, I will expose bugs that
> may be present at a deeper level (thus, having an impact on various
> APIs and the UI). If I can expose old API bugs that are restricted to
> this API, this is an unintended - but welcome - side effect.
>
> The one change I made to the old API was to fix a bug that I already
> knew about. I realized that tests would have exposed it much earlier
> and thus decided to write more tests.
>
> D.
>
> On Fri, Jan 8, 2010 at 6:15 PM, Ethan Jewett <esjewett@gmail.com> wrote:
>> I'm jumping for joy about the tests. :-) More failing tests and Jira
>> items means more bugs that we *know* about. But the same number of
>> bugs overall. So the more bugs we find and write tests for, the
>> happier I am! :-)
>>
>> So, one question I had here based on your latest commit, was if we are
>> fixing old bugs in the old API. My impression was that your bug fix
>> was to fix a new bug, and the un-noticed appearance of this bug
>> motivated you to start writing tests?
>>
>> Ethan
>>
>> On Fri, Jan 8, 2010 at 9:13 AM, Richard Hirsch <hirsch.dick@gmail.com> wrote:
>>> I wouldn't be jumping for joy yet.
>>>
>>> I love negative tests / finding bugs = lots of Jira items :->
>>>
>>> D.
>>>
>>> On Fri, Jan 8, 2010 at 2:56 PM, Ethan Jewett <esjewett@gmail.com> wrote:
>>>> Yay! Excellent!
>>>>
>>>> On Fri, Jan 8, 2010 at 1:58 AM, Richard Hirsch <hirsch.dick@gmail.com>
wrote:
>>>>> I've created a new Jira item
>>>>> (https://issues.apache.org/jira/browse/ESME-153) for testing the old
>>>>> API. I've also committed an initial set of tests. As we move towards
>>>>> our first release, we have to make sure that the core functionality is
>>>>> stable.  We've never really tested  possible inputs (puncution,
>>>>> numbers, etc) for actions, tracks, etc and I'm now using the old API
>>>>> to do this. This will also assure that "legacy" apps that use this API
>>>>> wouldn't be broken by code changes.
>>>>>
>>>>> D.
>>>>>
>>>>
>>>
>>
>

Mime
View raw message