[Re: Response to your last code review and FLUID-3116, FLUID-3117]

[Laurel A. Williams]

(reposted to the list - accidentally sent this directly to Michelle and 
forgot to cc the list)

Good to have your thoughts Michelle and I think you have solidified my

feelings on this too. There may be a little more I can do to ensure that 
the tests don't get broken as easily. For instance, I should probably 
add a warning comment to the data, reminding coders that changing the 
data may impact the tests. I was also thinking that the constants for 
the module indexes might be better placed next to the data itself, 
rather than in the tests, so that if someone adds or removes a module, 
the index values could be changed more readily.

Will await Justin's comments but am thinking that your reasoning may win 
him over ;)


Also, on another note, you and I chatted way back about other things I 
should change in builder.js. I put comment notes in the file to remind 
me (line 23) - but now I can't remember for the life of me what you or I 
meant, and Justin wasn't sure either from a quick read. So maybe we 
could talk about that quickly today?? I suggest if you have time to 
check the code out and look at it briefly maybe it will jog your memory 
and then we can talk later (maybe 2:30'ish) in the channel or if you are 
really busy a quick email would be fine too. Recall that builder.js is 
the "demo" code that a developer would add to their web page to use the 
customBuild component on their website. The comments you made were about 
moving some code from builder.js into the component.

Sorry for the last minute request...I only just realized I might want to 
check in with you before you went away!

Laurel

michelle.dsouza at utoronto.ca wrote:
> Hi Laurel,
>
> Good job on refactoring these tests. Please see my answer below.
>
>
> Quoting "Laurel A. Williams" <laurel.williams at utoronto.ca>:
>
>
>> As I was working with the tests, I noted that we reproduced the exact
>> same data in the tests. I refactored to use only one copy of the data,
>> but perhaps it was a case of "premature optimization".
>>
>> Note though, since this is a piece of demo code, the
>> fluid.customBuild.demo.completeFluidInfusionData data does not need to
>> change - it could stay the same even as infusion expands, since the
>> data for the 'live customBuilder' will always come from the
>> build.properties file and not from this data structure. However, I'm
>> thinking that people will be tempted to change the data as the real
>> infusion expands, in which case they will have to change the tests to
>> reflect those changes...and some of the hard coding will be a pain to
>> change.
>>
>> I did try to address this somewhat by providing some constants for the
>> indexes of the specific modules in the data which you wrote tests
>> around. In customBuild-tests.js you can find:
>>
>>        var INLINE_EDIT_INDEX = 7;
>>        var PAGER_INDEX = 8;
>>        var PROGRESS_INDEX = 9;
>>        var REORDERER_INDEX = 10;
>>        var UIOPTIONS_INDEX = 12;
>>
>> So, I think that outlines all of the details around this data
>> structure...what do you think...should it simply be duplicated in both
>> the tests and the demo to prevent people from breaking the tests
>> inadvertently? I'm not convinced, but honestly don't think it is
>> critical one way or the other.
>>
>
> I think it would be best to have a single copy of the demo data and 
> have the tests run against that. It's true that if someone updates the 
> demo data the tests may break, however, that might be a good thing. If 
> we at some point change the structure of the data, not just the 
> content, we would want the tests to break. It is more likely that 
> someone will update the demo data rather then the test data in this case.
>
> Also, as you said, I don't feel there is any need to update the demo 
> data when new modules are added. Unless of course the pattern of the 
> additional module is different - a special case. In that case, we'd 
> want the test data updated anyway because we would need to write a new 
> test.
>
> As someone coming into these tests cold, I actually like the named 
> indexes because it makes it clear to me in the test what is being 
> tested. It's easier for me to look in the data for 'pager' rather then 
> index 8.
>
> Hope this helps,
>
> Michelle
>
>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: laurel_williams.vcf
Type: text/x-vcard
Size: 279 bytes
Desc: not available
URL: <https://lists.idrc.ocad.ca/pipermail/fluid-work/attachments/20090821/4f72b6b9/attachment.vcf>