Closed OliverJAsh closed 11 years ago
Right, I knew this would come up and I think it's a very good thing that it did. Although arguing in the manner it was on Twitter was a waste of time, get it straight to Github! :dancer:
All the npm module is is a wrapper around git clone
- you can see this, the source is only about 10 lines! If people don't want to use it there's certainly no obligation to.
For the next three issues, we need to decide what level of "developer" we're aiming at here. For me, it boils down to this:
I also think a lot of the following is solved by having good documentation. I know this was pointed out on Twitter, and it does need work on, but right now when the project is going through rapid iteration we don't want to dive into docs just yet as they will need updating all the time. Very soon it will settle down.
I mentioned to @OliverJAsh last night that I didn't like how complex the JS code currently is - it is very overwhelming. I don't like how modules don't return themselves, as that's how RequireJS should work. I think we need to think about this more.
A tutorial on using Bower + one on how to remove / replace is needed. It's really easy to remove Bower.
Again, tutorials required, but I think having Grunt in there is a good thing.
Got to dash, but I will add more to this later, and everyone else who is reading this, please do!
Edit:
Got into work so can add a couple more comments.
Is it worth offering a custom build option on the site, like HTML5BP itself?
Yeah I wonder if that's a good solution. I know @OliverJAsh suggested multiple branches with different features, which I wasn't so keen on, but some form of build process might be the way forward.
Either way there definitely needs to be something that is basic for the 'generic user' that wants just the HTML/JS. Irrespective of whether you think people should know Bower/Grunt/RequireJS, a basic barebones one will suffice.
Also, if you think it's going to be more complex than H5BP, try not to tout it as something that is the JS equivalent of it - 'cause it's far from it as it stands. :)
So I had a think about this on the tube on the way home.
I really like the idea of some form of build tool, so you have the basic structure, which is without RequireJS, Grunt, etc, and is more like how @benhowdle89 first imagined it. Then perhaps you have a tool that you can run to generate custom versions.
heisenberg requirejs
heisenberg grunt requirejs
heisenberg test-jasmine
Or whatever. The base Heisenberg is like H5BP but there's also more available if people want it.
The technicalities of building such a tool might be a bit challenging but I think it could be interesting to do and also really useful.
Thoughts pls!
Fully agree with this. God knows how to build a custom build tool.
My vision for Heisenberg admittedly was aimed at people who weren't really that comfortable with Bower, Require and Grunt etc, but I think it'd be awesome to have a build tool, ie. something you could run on the command line and it would create a 'dist' folder with concat + minified files in there.
I think you need to steer clear of forcing things to the command line. What's wrong with just a simple zip download of everything that's required, or a simplified branch, or something?
Why is everything becoming about the command line recently?
I think you need to steer clear of forcing things to the command line. What's wrong with just a simple zip download of everything that's required, or a simplified branch, or something?
Nothing at all, and we'd offer that. For those who want a bit more, perhaps we could look into a build tool. Also, we've not decided anything yet - just discussing!
Reverted to a more simple approach to base Heisenberg. For future plans see #46.
There have been issues raised about the accessibility of Heisenberg in its current state.
Since its initial introduction, we have added RequireJS, Bower, and Grunt. However, to stay true to this project's goals, we still need to cater for users who:
I have thoughts on all of these. Some of them we are obliged to fulfil, and some need to be protected to ensure we continue to support the notion of "delete key" friendliness.
I'll follow up with more of my thoughts on how we can improve accessibility with this criteria in mind. In the interim, feel free to comment with any ideas. :smiley: