Skip to contentSkip to conversation

Repository Growth

DISCUSSION [email protected] 5 participants 5 messages Sep 19, 2018

lore
Share & tools
lore
#0 lore raw
Quoted 1 line

Thoughts?

I had the same idea from reading through the presentations recently linked to on this mailing list.

OpenBMC lists its features on its front page
(https://github.com/openbmc/openbmc README) and suggests there is more information in https://github.com/openbmc/docs. However, the information is not structured as well as it could be for someone to learn about each feature. I propose to list each feature along with the following information (as applicable):
- reference an external website or standard
- categorize: host-facing, management network facing, application
- explain terminology (give synonyms)
- explain 1 or 2 use cases
- link to the docs (which should explain customizations)
- link to the source code repo

I would start with the features listed in openbmc/README, and add features (or interfaces?) from James's presentation:
- Host interfaces: KCS, BT, mbox, IPMB, USB, VGA
- Network Interfaces: Redfish, https, ssh, rmcp+
- Applications: EWS, KVM, IPMI, FSC, SEL, SOL, Chassis Control, ASD,
Firmware Update, Configuration Manager, Sensor Monitor

So, the proposal is to group repos by feature and interface.

#1 lore raw
Quoted 72 lines · On Wed, Sep 19, 2018 at 12:56 PM <[email protected]> wrote:

Date: Wed, 19 Sep 2018 09:42:16 -0700
From: Patrick Venture <[email protected]>
To: [email protected]
Cc: OpenBMC Maillist <[email protected]>, Brad Bishop <[email protected]>, Nancy Yuen <[email protected]>, [email protected]
Subject: Re: Repository Growth
Message-ID:
<CAO=notwvBca4ANU6b-kQggP53cM2CdsCVV5QqsBYUCJ_jB+Xkg@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"

Quoted 11 lines · Adriana Kobylak
Quoted 9 lines · Patrick Venture

With the growth of number of repositories, it might be overwhelming to someone who wants to set things up.

I'd like to write up a directory that groups the repositories by purpose. I wasn't sure if this made sense, so I wanted early feedback before I started. The maintenance of it would be horrible for one person, but if someone wants their repository to have details and so on they can add it.

Quoted 6 lines
Quoted 4 lines

I think it's important or useful to know what's out there and how things work together or don't, and what options there are for developers.

+1 . Do you have a template in mind on how this could look like?

I've been toying around with some ideas on how to organize the information. Grouping the subtrees together is a no-brainer, but many daemons don't really fit into a category. Maybe that's fine, maybe then it'd be uncategorized.

I was imagining:

{repository name}
{short description}
works with: {list of related pieces}

for instance:
"""
### phosphor-hwmon
phosphor-hwmon periodically reads and reports hwmon sensor values to dbus. works with: phosphor-host-ipmid
"""

You don't need to run phosphor-host-ipmid to use phosphor-hwmon, and phosphor-pid-control and other daemons can also listen to sensor values or read them over dbus, so I don't know how complete such a list would need to be. So maybe the list should be less explicit. I realize that the recipes themselves reveal the true dependencies, either virtual or specific and I don't want to repost information. I just want to present a better interface to the set of repositories and how they work together (or don't) than github has.

Quoted 2 lines
Quoted 2 lines

Thoughts?

I had the same idea from reading through the presentations recently linked to on this mailing list.

OpenBMC lists its features on its front page (https://github.com/openbmc/openbmc README) and suggests there is more information in https://github.com/openbmc/docs. However, the information is not structured as well as it could be for someone to learn about each feature. I propose to list each feature along with the following information (as applicable):
- reference an external website or standard
- categorize: host-facing, management network facing, application
- explain terminology (give synonyms)
- explain 1 or 2 use cases
- link to the docs (which should explain customizations)
- link to the source code repo

Quoted 4 lines

I would start with the features listed in openbmc/README, and add features (or interfaces?) from James's presentation:
- Host interfaces: KCS, BT, mbox, IPMB, USB, VGA
- Network Interfaces: Redfish, https, ssh, rmcp+
- Applications: EWS, KVM, IPMI, FSC, SEL, SOL, Chassis Control, ASD, Firmware Update, Configuration Manager, Sensor Monitor

Makes sense to me.

Per the mailing list discussion yesterday on the hackathon, I'm also starting to work on some "digging into openbmc coding" type docs. That will probably deserve it's own directory?
- Getting Started: Dev env setup, hello world via SDK/QEMU, adding a new system/layer, customizing webui and other items for new system, bitbake using local repos, ...

Quoted 7 lines

So, the proposal is to group repos by feature and interface.

Quoted 3 lines
Quoted 2 lines

Patrick

#2 lore raw
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

The maintenance of it would be horrible for one
person, but if someone wants their repository to have details and so on they can add it.

Is it possible to scrape some of this info from github, gerrit, etc.? Perhaps portions of this could be automated to help with maintenance and to keep the information up-to-date.

#3 lore raw

+1, I like the idea of having some form of automation to help with the maintenance of this information.

----------
Nancy

Quoted 10 lines · Jason M

The maintenance of it would be horrible for one

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

person, but if someone wants their repository to have details and so on they can add it.

Is it possible to scrape some of this info from github, gerrit, etc.? Perhaps portions of this could be automated to help with maintenance and to keep the information up-to-date.

#4 lore raw
Quoted 15 lines · Nancy Yuen

+1, I like the idea of having some form of automation to help with the maintenance of this information.

----------
Nancy

Quoted 8 lines · Jason M
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

The maintenance of it would be horrible for one
person, but if someone wants their repository to have details and so on they can add it.

Is it possible to scrape some of this info from github, gerrit, etc.? Perhaps portions of this could be automated to help with maintenance and to keep the information up-to-date.

I think this could be scraped if the READMEs were formatted in a machine-readable way consistently.

Quoted 2 lines