Sprint E

[jbenet]

Sprint Goals

1. Wire protocol + interop with node
2. Make gateways faster / better
3. Perf + Bugfixes

Sprint Participants - @jbenet @dylanPowers @whyrusleeping @travisperson @wking @cryptix

Sprint Deliverables

• make it easy to work with gateways
  • remove docker, init script, auto-rebooting (@jbenet)
  • cpu + mem profiling on gateways (@whyrusleeping)
  • change nginx to do fallback (not round robin) (@jbenet)
• perf/bugs:
  • change the memory profiling (@whyrusleeping)
  • save memory profile + binary to ipfs (@whyrusleeping)
  • scope out how testing will be done on both webui and api (@travisperson, @jbenet)
  • /files with thousands of files added makes my browser hang ipfs-webui#20 (@travisperson)
• Specs / scoping out
  • DHT spec (@jbenet)
  • Keystore spec (@jbenet)
  • scoping out how signed binaries + autoupdate will work (@jbenet)
• interop with node
  • Protocol wire spec (@jbenet)
  • swapping out the stream muxer (or having multiplex in go) (@jbenet)
  • protocol-stream (@jbenet)
  • option (or build) to disable encryption for interop with node (@whyrusleeping)
• Community
  • Waffle documentation (@dylanPowers) Document waffle.io/ipfs/ipfs infra#8
  • Explain the label scheme (@dylanPowers) Explain label scheme used community#10
• make all adds add to an object representing local state
• scope out init + daemon sprint (@jbenet)
• reimplement goprocess without mem leak (@whyrusleeping)
• Address memory leak (likely culprit is goprocess)
• json encoding for add (@travisperson)

[jbenet]

@wking + @krl -- still need to add things for you this week-- chat briefly on irc?

[wking]

On Mon, Apr 27, 2015 at 01:25:46PM -0700, Juan Batiz-Benet wrote:

@wking + @krl -- still need to add things for you this week…

[ ] organize filesystem node creation and extraction in shell/fsnode
and core/fsnode, deprecating the existing importer and
core/coreunix interfaces and migrating the internal use to the new
interfaces. I may only get through the addition commands this
week (WIP in ipfs/kubo#1151. This is part of a larger push to
clean up the existing file handling before working on access modes
(more discussion in ipfs/kubo#846).

I'd like to open an issue to figure out our migration strategy. Do we
need ‘gofix’ style translators? Should we be logging runtime
deprecation warnings? Some folks do some sed-like things with ‘go
fmt’. Maybe we're just explaining migrations in our release notes?
Anyhow, it would be good to have a plan written up somewhere, but I'm
not sure where that place is. Ideas?

[whyrusleeping]

@wking deprecating importer? in favor of what?

In my perfect awesome world where I wrote all the code and spent lots of time making the interfaces perfect, We would have the importer, with all its knobs and switches on the low end (probably), and coreunix.Add being all nice and user friendly on top, and then the commands.Add would just call into coreunix.Add.

I guess the further refinement of that idea is that all commands are just UX wrappers over the core* methods, and the core* methods handle a bunch of logic in order to make calls to the low level importer style, raw methods.

maybe we should rename core -> mantle, and call the lowest level type API commands the core...

Maybe i'm crazy, @jbenet @wking thoughts on this?

[jbenet]

@travisperson for "scope out how testing will be done on both webui and api" if you can propose a system or two i can review it and suggest changes. maybe in: ipfs/ipfs-webui#38

[whyrusleeping]

potential issues to address next sprint:
ipfs/kubo#1082
ipfs/kubo#1050
ipfs/kubo#964
ipfs/kubo#1179
ipfs/kubo#1150

In general, try to do something with ipfs, record difficulties, try and fix them

[whyrusleeping]

also, if anyone is looking to poke the webui:
ipfs/ipfs-webui#41
ipfs/ipfs-webui#42
https://github.com/ipfs/webui/issues/43
ipfs/ipfs-webui#38

[whyrusleeping]

I would also like to get more testing around ipns and path resolution written, testing various commands with different path syntax (/ipfs/ vs vs /ipns/)

[whyrusleeping]

and this jbenet/go-peerstream#5 also needs to get done