Re: Feature test regression failures

On Mon, Mar 27, 2017 at 12:05 PM, Atira Odhner <aodhner@pivotal.io> wrote:
>> A change as large as moving to React will be for v2.x, not v1.x, so we
>> cannot really wait unless we want to write off the feature tests as
>> unusable for the forseeable future.
>
>
> I don't think we should wait on moving the entire codebase to react before
> releasing features that use it. React allows us to implement things
> piecemeal so for example we can implement this tree in react and ship that
> as a fully functional user-facing change. Over time we can introduce more
> components in react while still having a working product.

Maybe - I'd need to see a viable patch with minimal impact first though.

We've already broken many of the normal rules we play by with pgAdmin
4 by cutting releases from the master branch rather than keeping a
stable branch with only bug fixes in it, and I'm extremely wary about
shipping largely non-essential changes in a minor update without being
able to do the level of QA we would for a major release. The risk to
the user and to our reputation is too high.

>> I'm tinkering with ideas for optimising the code we ship to users in
>> my spare time. This isn't something I'm prioritising. If you're
>> working on similar ideas, please do share.
>
>
> The best practice for using React is to use a preprocessing pipeline so
> we'll definitely share that as we get it set up. We will definitely be doing
> minification and bundling code into a single file.

That'll only work to a point. What happens if the user installs a
plugin to get some additional functionality? We still need the ability
to ensure code can be added to an existing installation.

--
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake

EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company