Information architecture of site

[mor1]

Following discussion of @dbuenzli feedback:

• Rather than recommend building mirage-www as part of tutorial, perhaps building a personal site would be more engaging?
• We should not hide how to report issues under "Contributing" but provide an explicit "Troubleshooting" page that describes the common pitfalls, how to solve them and then provides guidelines on how/where to report an issue would it persist.
• We should remove the "library" links on that page as they obfuscate Mirage specific information (e.g., regexpes, COW or htcaml).
• The Meeting Minutes are a couple of months out of date (latest showing is 2015-12-02).

Generally it might time to look again at the flow through the site...

[dbuenzli]

Rather than recommend building mirage-www as part of tutorial, perhaps building a personal site would be more engaging?

An http hello world over raw tcp/ip (no cohttp) should come first I think. Then I would branch this into diverse things, e.g. going down to simplified (if too complex) mindyesque networking demos or up to make a toy webservice or serve an http page using a higher-level http library. The idea here is that the examples should reveal the structure and different layers of interaction with the mirage.

Also I think there should be a few toplevel sections about important aspects of the life cycle of an unikernel (something like developement, debugging, profiling, deploying).

There should also be a clearer separation about installing mirage, reporting problems for it etc. and unikernel development itself.

[hannesm]

Related to this is the thread on mirageos-devel in September 2015 about "Dispatching boilerplate and the tutorial"

[mattgray]

An http hello world over raw tcp/ip (no cohttp) should come first I think. Then I would branch this into diverse things, e.g. going down to simplified (if too complex) mindyesque networking demos or up to make a toy webservice or serve an http page using a higher-level http library. The idea here is that the examples should reveal the structure and different layers of interaction with the mirage.

That's what I tried to do with the "Intro to Mirage" session I did at Devwinter in Cambridge in January. Started with writing HTML using raw tcp/ip, and then introducing cohttp, KV_RO and randomness, to illustrate the idea that Mirage is a library OS. Example code for that is at https://github.com/mattgray/devwinter2016/tree/master/cat_server (not using Mirage 2.7 though, so no Functoria just yet)

[mor1]

Hi @mattgray -- noticed this and thought I'd have a stab at Functoriaising it -- PR in mattgray/devwinter2016#2 in case it's any use

[mattgray]

superb, looks great @mor1 ! I'll have a closer look when i'm done at work :)