BADCamp 2018: So close to home for me and so much more walking than other events

BADCamp logo over a city skyline of San Francisco

I go to a lot of camps, but very few camps come to me. Bay Area Drupal Camp (BADCamp) is in my area, but over in Berzerkeley. It is beautiful college town that is just over the bay, only 45 minutes door to door. The last bit from the BART train to the UC Berkeley Student Union is not a subtle elevation change. But a good uphill walk wakes you up and get the blood a pumping which put us in elevated moods as we got together to learn and celebrate the community at BADCamp 2018.

This year’s theme was the circus! The decorations were all circus-y and there was cotton candy and popcorn abundantly flowing. And the juggling! There was a lot of juggling.

Food and Fun

Wednesday

This was the first year there was a Wednesday for this event, having shifted form a ‘Thursday to Sunday’ schedule to ‘Wednesday to Saturday’ one. Coffee at Berkeley is not better or worse than any other University of California. It is pretty consistent and average. As an attendee I can’t complain about the price point of free. Pastries were also abundant.

There is no lunch provided at BADCamp. Since this is a free event, it would be surprising if lunch was made available. They just opened a Veggie Grill in Berkeley and I dined there with a colleague. No tweet of mine for that, so here is one of theirs.

Also of note, I went to see Jawbreaker on Wednesday night. A couple people from camp went as well.

Thursday

More coffee, more pastries and more summits, trainings and hallway conversations at the camp. For lunch I grabbed some Thai food at my favorite Thai place in the area, Thai Basil. Sometimes the greasy spoon places are the best.

After Party

Thursday night we had the first of the official after parties. We gathered at the rather swank Tupper and Reed for some free drinks courtesy of DDEV who threw the DevOps After Dark Happy Hour. Many a good conversation was had.

Having a couple drinks a few of us wanted to get some amazing plant based foods. Berkeley is home to the best vegan deli I have been to personally, The Butcher’s Son. I had the best giant bacon patty melt I think you can possibly put together. We all left completely stuffed.

DrupalCrawl

A Drupal event where the awesome Jason Mickela is present will almost certainly involve a quest to find microbrews and specialty ale under the banner of DrupalCrawl.

DrupalKaraoke

One of my greatest joys in this space is getting together to sing as a group. An apex part of this aspect of my life is getting to know JD Flynn and Chris Weber better. JD actually did the footwork on getting people excited and pushed to the venue of choice for this event Nick’s Lounge. I was overjoyed that the Kanopi Studios team was already there when I showed up and we all sang together through the rest of the night!

Friday

The first day of sessions at BADCamp bring with it waffles. Not just waffles, but fresh made waffles with a lot of fixin’s. Coffee and tea flowed as well. I went back to Thai Basil again for lunch. It is good.

After Party

BADCAmp is known for a few things. Throwing a very good after party is one of those things. This year the Platform.sh team, which I learned is properly pronounced “Platform, Shhhhhhh”, threw the event. We got to go on the roof of the Marsh Art Theatre. We got to dance to some thumping tunes and get our faces painted. And of course there was a photo booth to show how we were affected by the provided drink tickets. All in all a great time for hanging with the community.

After After Party

In the tradition of this event, some of us went out late into the night and made a pop up kitchen happen in the middle of nowhere, complete with disco lights and carpeted areas under the stars. It was a blast.

Saturday

More waffles and more coffee. I personally got some cotton candy for that sugar energy to get me through the day.
Lunch was again Basil Thai, though I had almost ordered ramen next door. That super hot chili and peppers though. Seriously if you are in the university area of Berkeley around lunch time, go give these nice people some business.

Lunch related tweet:

After

Thanks to Avi I got to go see Middleditch and Schwartz on Saturday night. It was a very great way to end a very busy but great week. Even got to sneak in some Karaoke before bed as the official end to my BADCamp.

Sessions

Thursday DevOps Summit

How Do I Host Hundreds of Drupal Sites?

This is not your normal set of sessions. The Summit is to encourage community discussion of the best ways to approach problems. FOr a number of reasons with setup and my schedule, I only got to attend one of these summit panels, but wow I am glad I did. The chaotic notes that follow are my best attempt to capture what happened, sorry if you get lost. Basically, running a lot of sites is a challenge that has multiple approaches. The conversation is far from over though and this is the type of thing I love talking about if you see me at another event.

Raw Notes:
Panel: How Do I Host Hundreds of Drupal Sites?
Shawn Dearmond: University of California SiteFarm – on campus
Thom Toogood: GovCMS (Lagoon/Kubernetes)
Daniel García Mont: ASU WebSpark
John Bickar: Stanford Sites –
John Grubb: Platform.sh – not in attendance, never caught name of Platform engineer.
John B – like a 1000 sites. migrating to Acquia
Shawn – 300 Drupal sites in the group, another 300-400 on Drupal sites or clones, maybe 150-200 random sites people ‘just run’
Thom – amaziee.io, Australian government – 10-12 in next 2 weeks
many more launching soon, many each week launching
federal state and local
2 sides of the program custom and multisite
John grub not attended – Platform.sh -how to manage a lot of sites, not all Drupal
Daniel García – close to 2000 sites, all over the place, all tech
last few years sunk into cost recovery model
380 production sites on Pantheon
Drupal distro – 77% on the Pantheon sites
others
Mostly in Higher ed for this summit, but all over the place for this panel
Why the platform you picked?
was it costs?
John B 17 people
originally built service on premise
sites need to be created without
tricky to try and automate that
need something to spin up all the services needed for the sites
Looked at site factory platform
robust API
LDAP to Site Factory
log in and all the things
Shawn – similar to Stanford
Started on Pantheon – ended up with a Custom upstream
very bespoke
as started doing this over and over again
building out distribution to solve the core use case
300 live sites on site factory
in dev will be live soon
individual departments
grouping functionality with sub groups
works well with the hierarchy of a college
see sites within the department
did need and orchestration system
see all the sites at once
4 hours to
building recharge work on Pantheon and some good old site building on factory
no cost to campus customers
Kubernetes
few other competitors in the DLS in the city
Gov design higher demand scale up and scale down
high availability compelling automated testing around containers
very simply each branch is a full build
Gitlab CI
change model version change
QA and supervisory model
for government, didn’t want anything in production cluster that aren’t approved Gitlab provided validation structure
very flexible
wanted to maintain some control
– Platform is specific unlike these other use cases
composer install run for you
just update your local and bring new branch to Platform
clone original branch and create a new env to get live services
apply new updates and run tests in new environment
if like the tests, just merge it in and that is deployment
Replicate-able for multiple sites
can push to hundreds of remotes
relatively simple using scripting for coordination layer
fleets of websites
ASU – picked Pantheon about 5 years ago
Maybe a little different
this is a different model
service partnership with other departments
contract with them to build
Pantheon was willing to work with them
Good dev experience on-boarding process was more flexible
it is a pull model – CU available
hybrid central and decentralized
websites owners must update sites
deploy updates fro service partners
9 of the team are devs
cost recovery in 2 areas
service and hosting
2008 the leadership introduced the model legacy hosting free hosting offering
no way to see what was abandoned
Faculty does not want to pay for their websites
departments pay for services though
Comments of picking a new or many new platforms
one of the things we have seen, move to standardization
looking for 80% of sites in one place
easier to roll out updates on app layer
John B – open env, anyone on anything in CGI
on IT for updates and they didn’t get done from app end
no life cycle management
each site has own doc root each on git branch
updates on infrastructure
giant multisite
Useful Site factory, lot of updates quickly is easy
one branch to push to ll sites without much intervention
Pushed updates to many sites in few hours with Drupalgeddon
Shawn – several different processes
Drupal 8 stuff much easier to use
CircleCI
Site factory process
Bitbucket for code
proboCI to do builds and run through 2000 Behat scenarios on fresh sites
not updated sites
moving to updated sites
PR merged, pushed to Acquia and runs their pipeline
saves to branch in their Git repo
tag to a release number and push, chugs through and updates fast
takes a few hours but with Behat takes about 20 minutes to run
less than a hour to do all changes on their end, then a few for Site Factory
On Pantheon site by site basis, takes some time
have to hit every site
some on A vs P
Recharge work for custom sites
‘free’ Acquia sites, but it is locked down
if you want to write your own, other product, Pantheon CU
you pay for that though, full thing
extra costs for maintenance
even have export from Acquia to Pantheon process
no PCI or HIPPA or other specs on site factory
lets them be flexible
“Who gets what module in” is a sticking point with Multisite
Thom – how do you avoid bloated MS setup
hard not to
with D8 focus, trying to get profile as small as possible
biggest change is output from profile as a make file
also outputs Docker images now
test these images and host themselves
extend and customize from those images
all can really do is change the Theme folder, not even managing a D codebase
all built and tested Upstream
building the project on the latest image
added a test image
run their own image
base themes and modules, lightweight controlled
Security certification that they
PLatform- updating and just updating composer lock and push it doing it wrong
all module updates and migrations
updating the infrastructure – growing team doing it
hosting your cluster built on read only containers
once built it is left alone
all changes get a new container
more layers to this. Regions and container versions can vary
problems with Drupalgeddon solve in unique ways
web application firewall in the regions
update firewalls for all clients first
ASU – build process 300 production sites
very greatly with each site
custom modules to standard sites
volunteers in the web community ot test things
merge after testing, pull model, each unit responsible to get to production sites
aim for monthly, but that fatigued people
now quarterly , 24 hour turn around just for the update setup
UCDavis – Using the test env on Site Factory, email sent to folks ahead of time
Division between special customizations
Last Q
Remaining challenges for the current platform
– 800 sites migrating into site factory
it is a lift and shift
legacy Auth with saml
latest core but getting to latest contrib modules is a challenge
Davis – biggest issue is difference between environments
test past fine on local and then fails on Probio
and different than prod
containers should be solving this
Acquia’s infrastructure is
Pantheon is not much better
caching layers is difference between local and CI and production
they spend more time fixing than making new code
i/o same set of issues, but future is standard containers everywhere
new thing for most people, training is a real concern
150 devs to onboard
Platform – user tools – HA solved, Various env solved
Didn’t solve deep learning curve at beginning
lot of issues around users understanding Platform’s opinions
Fleet management is automate-able but not automating it out of box
amazing tools you can use, Lando and DDEv, but noting direct for them
ASU – most users don’t need to worry about these things in ideal world
but not real right now
RFP is to get to something that automates

Friday

TALK NERDY TO ME: TRANSLATING TECHNICAL JARGON INTO PLAIN SPEAK FOR THE REST OF US
LEIGH ANDERSON

Communication is tricky and it is very easy to forget that not everyone is always on the same page as far as terminology and points of reference. This was a great talk that broke down how communication loops occur and how we can use a pretty simple process to make sure we are communicating well with people in all sorts of roles and skill levels. I know I am going to be reviewing my communications better into “what does it do” and “why” as I present new ideas moving ahead.

Raw Notes:
She translates a lot of technical solutions into plain speak for clients
worked at the Drupal Association, talking human side of Drupal
how it had changed their lives
Communication is
encoding – the sender packages it up to send
then sending over channel, letter, song, dance,
Message Decoded
reply feedback encoded
channel sent
decoded feedback
Barriers happen all the time at all channels
maybe technical, maybe personal, maybe cultural, attention, disability, etc
Using right words
with the right people
in the right medium
at the right time
for the right response
Reducing barriers
encoding/decoding problems
incomprehensible jargon
lack of shared knowledge base
vocab barriers
cultural values difference
Scheme – US English means plot, underhanded, UK- just a plan for an org
Issues with the medium
hearing, vision or other disabilities
physical impairment
inherent encoding flaws
Communication breakdown example
“what are you thinking for dinner”
meant ‘should we get pizza’ gets ‘make your own damn dinner, I don’t wanna cook”
well that escalated quickly
if you pause and realize there is a barrier there, maybe should ask or clarify
best version, sender is much clearer “I crave pizza, can I order one or are you planning something else?”
Know your audience
How can I phrase this in terms this person will understand?
how can I talk about this in a way that will make them care?
DO NOT be condescending
What are they doing now?
What are they trying to do?
What’s keeping them from doing the thing?
What do they value?
What makes them successful?
What makes them fail?
And find their barriers
Same background?
Same native language
Is this best medium?
is message structured the best way for this medium
example
Non-profit living history museum in Austin, TX
volunteer org
I spend a lot of time sharing info about bookings to people.
What is he doing now?
What is he trying to do?
What is keeping him from doing it?
What does he value?
What makes him successful?
What makes him fail?
Prefers spoken communication
talk about how solution saves time, and makes his life better.
Not technical answers
Why?
Find clarity with your inner 4 year old
What’s the point?
ask why until you have the most basic answer
How questions is a rabbit hole
vs what does the thing do
Avoid technical language and find concrete real world terms
example of cache solution
broken down to ‘makes the site faster’ and not delving into tech jargon
Might seems like a long process and lot of work
at first it is
then second nature
then improves all communication
most people will meet you half way
clarity and kindness makes things better
https://sidewaysdictionary.com – analogies explaining tech.
internet linguistics
gretchenmcculloch.com
skillsyouneed.com/ips/communication-skills.html

GATSBY AND THE JAMSTACK: AN INTRODUCTION FOR THE DRUPAL/LAMP MINDED
NICK LEWIS

JAMStack is the now and future for the front end. We are moving into specializing the CMS to just worry about storage and access to the content, moving to things like Gutenberg for editing experience and microservices for, well, services like forms. Gatsby is literally helping people make sites that are ridiculously fast. My favorite line from this talks was “The Gatsby team tries to make it hard to build a slow website.” This is a great intro and the example repo is very practical and usable out of the box.

Raw Notes:
He does not like JS frameworks
got burned on projects
Drupal learning is hard, really hard for JS template builder to understand
I’ve seen this before and its the work of the devil
now believer
JAMStack is a lazy term
secret sauce is sheer hosting laziness
production is flat files only
interactivity is 100% client side
Content is managed via git or a CMS on a separate server
Buzzwords but practical
$5k a month to $5 a month
removes all tuning and monitoring needs
very secure
CMS-> Gatsby -> JAMStack -> CDN
Code samples
how it is laid out
Gatsby is like a compiler
much simpler structure than CMS
index.js
not everything loads automatically, you have to import things you need on each page
Helmet instead of head
HTML rendering via JS
needs to be compiled but Gatsby takes care of it
some syntax to learn
Define a new react component
passing in Props variable
children – lets you put components in components
compared to Drupal, this is simple
Style components
defining style on same page as component
keep things together
instead of giant SASS directory, just in one place
article teaser styles that react to that
renders in CSS properly
Setup
Drupal side
JSON API – in core in 8.7
have to install right now though
make anon requests or http auth, protect data
webhook with something like Netlify
site never goes down on Netlify
set up a build hook
API to let it know it needs to rebuild the static site
GraphQL – it is amazing
get all pages and articles
copy/paste basis
taking initial page query and passing into a template
but after the query is written it is familiar and simple
images are tricky at first as well
Gatsby wants to make it hard to make a slow website
lazy loading is automatic and taken care of by platform
Kyle’s desire to mathematically save off milliseconds drives this
https://github.com/zivtech/gatsby-drupal-kit
no views integration, must parse this in

DESIGN FOR NON-DESIGNERS
ROXY KORANDA

One of the most important books I ever read was The Non-Designer’s Design Book, by Robin Williams. This was required reading for me and helped me learn to be bold and stop centering everything. Design is something we are all actively participating in, from end user to conceptual artist, we all feed back into the system. No matter what your role, understanding how to better communicate visually is a valuable skill and I am very grateful for this session. Also, learned I have been misusing the hyphen and the em dash — a lot.

Raw Notes:
75% of users admit to judging a company based on website design
opening a designing program is not the first step
what do you want to do and why?
Strategy Toolkit
What should your message be? Focus on goals
audience
voice and tone
baby shower vs metal show
typography needs to reflect the right to the right audience
General Advice
remember goals and audience
keep it simple
focus on content
be concise
consider context
KIS – easy to get carried away, don’t go overboard
Focus on the content
use your strengths, think of cleaver headline over bells and whistles
core model exercise
how will people arrive at this page?
don’t bloat, can you live without it
consider context
flyer or website?
flyer in a grocery store, what is the rest of store like? What are other flyers like?
what limitations around how to use color
most colors print darker and duller than on screen
make sure you have ample frame/space if cutting it yourself
Composition
Prioritize
get to main points first
sketching is a great way to land on ideas and iterate quickly
consider maybe a wireframe
content prioritization and consistent alignment
visually anchored
whitespace is needed for design
not wasted space
less opportunity for distraction
looks luxurious
Good use of whitespace increased reader comprehension by up to 20%
emphasis visual space
consider the whole system
Images
learn Photoshop or Gimp deeply
just go outside, natural light makes things look good
know if you have image skills going in,
some tricks are very simple like color overlay or blur
GIMP is very good free Photoshop
Pixlr is good but a step down
don’t get too crazy
want the darkest parts of photo near black and lightest near white
adding saturation to match real life
adjusting color balance
blue? add orange
images indoors always look too warm
cool outside
illustration and iconography
be smart and avoid clip art
avoid flaticon
The noun project – attribution
iconmonster
vector, scales very well
Logos
do not need to be huge
smaller is better a lot of the times
consider background color
keep in mind what backdrop they need to go on
fonts
Avoid bad ones
Google your Font and see what the world really thinks
Avoid wordart
in selecting fonts, Google fonts are great resources
mixing fonts can be dangerous
same font that is everywhere but has multiple weights
many weights makes it more flexible
Typewolf is a good resource for use in the wild
Type Hierarchy
big bold headings get points across
make sub-points same weight on a scale
avoid long line lengths
legibility
contrast to be clearly legible
Dashes
difference between hyphen, En Dash, Em Dash
– hyphen just –
– en Dash alt + –
— Em dash Shift + alt + –
Color
Palettes – stick to one or two if you don’t have a lot of resources
Coolers website, color palette selector suggester
but it does not tell you what colors work together
think of A11y, color blindness
colorvale
stark, color bind simulator for Chrome
Spectrum is an extension for Chrome that simulates color blindness
immerse yourself in good design and take notes
look at other work and borrow elements but do not copy
designspiration
dribble
slides online

BUILDING WEBSITES WITH REACT & GATSBY
Kyle Mathews

There were a few sessions about Gatsby at this event, but only this one was the original founder himself, Kyle. After a few equipment issues in the very packed room we were quickly looking at code and examples, which are kind of hard to take notes for, using my style anyhow. Good overview and thanks to Kevin Thull I think the recording is going to be a great way to see this.

Raw Notes:
Background in Drupal
Once React started he didn’t want to do anything else
he wanted a simple tool to make react easier
server render React to static HTML so sites load blazingly fast
But then load minimal JS
seamlessly add interactive/dynamic sections to the website as well
Write all templates in one format
Early on Gatsby was similar to Jekyll/Hugo and the like
people started using it and found bugs and made feature requests
How to integrate WP, COntentful or Drupal
What about themes, Code and data splitting and large sites?
Not just for Static sites,
Gatsby could solve problems similar ot what Drupal solves
Traditional CMS Presentation layer over COntent layer
Decoupled
Presentation layer – application layer
how do you bridge this gap?
plugins and GraphQL
Build time GraphQL layer
designsystems.com
justdoit.nike.com
edgedb.com
Rachio.com
codesandbox
how to get started…
link component
pre-fetching
demo start states
code code code
source plugins
code examples

WHY DOCUMENTATION MATTERS (EVEN WHEN IT SHOULDN’T)
MICHAEL NIELSON

The same way we are all designers, we are also all responsible for documentation. Even if you don’t code, it is very likely that when you need to use a product you will reach for those docs. No matter how you generate product there should be a good paper trail that explains how the thing works but also why it works. If we stop thinking of our reader as some random technical person and start writing to ourselves in the future, we are bound to create better, more readable and overall more enjoyable docs. This was a super solid presentation I hope every product builder sees.

Raw Notes:
We are all users no matter hat else
Docs are the first steps someone takes when they start out
good first impression point
On twitter @Codewisdom is good to follow
we are writing docs for our future selves
retroactive learning
can re-encode memory
step through creation process again but can easier codify why and what
being stuck sucks, getting unstuck feels amazing
writing docs in hopes that my struggles aren’t yours
Docs are good for a team
brain dump
billable hours
fewer meetings
mind state
fewer trivial issues
more brains grasp what’s going on in the code
reading the whole doc often gives you what you actually want
Good for ourselves, good for the team, good for the community
Why docs suck
hard to find, opaque and often out of date
writing code is NOT writing documentation
not even close
Makers aren’t users
reading is not understanding
some concepts to consider
imagine writing a doc for a 4 year old in making paper airplanes
or how to tie a bow tie
— no images allowed
We don’t know what we don’t know
Outside context problem
book: “Excession” – Iain M. Banks
poor language
translating requirements into code is hard enough
translating code back in to meaningful docs
packet dropping is and should be expected
not a software problem
useful knowledge transfer is a whole other field – Education
improving the concepts undergirding education
epistemology
We are not alone
written communication in inadequate to express some concepts, like care crashes
code has advantages
our language is simpler
concepts less nuanced
convincing others on why this is important
Getting buy in from the team
especially stakeholders
can’t be underestimated
think in long term shift
uses logic, self interest (SS)
long term
BUS plan
what happens when so and so gets hit by a bus
pros – extremely visceral
highly motivational
can use real names
downsides
one time use only,
short duration of effectiveness
management by crisis
average time to first commit? ON average 6-12 hours
huge giant welcome mat
easier for people to get started and easy to promote to your friends
lower training costs
no need to fly folks in to train them
save on many fronts
Documentation fun?
fun to consume vs fun to create
for one sprint, do nothing but write docs
not one line of code
get the whole team involved
get food and drinks, hourly stretch breaks
get your PM to review
do mini sprints
60 minutes
team on site, a couple doc sprints
add paragraph breaks
add Flash, not the tech
images
gifs
emojis
add photos
tools for multimedia
licecap – make a gif
quicktime – audio and video
create doc stories
break into chunks
assign it to someone
go through approval
ship it
creating an Index
automate it, if you can
if you can’t make it single page and exhaustive
non-technical foks can spot Jargon faster
refactor docs
add a todo at least
encourage new hires to contrib to docs
better and better
requires revision
maybe somethings changed
mind the gap
what is trivial conclusion to senior developer may be a curveball to someone new
focus on logical jumps that appear to be simple
sourcing bashrc for instance
docs as acceptance criteria, mandate docs creation as part of marking off something as done
maybe docs as first pass
don’t do it – anti-patterns
not documenting, this is the worst
actually having bad docs is worse
if out of date Mark it as such
make a half assed effort, ehh, it does stuff
don’t make it easy to find
make it top level
Don’t violate Wheaton’s law – don’t be a dick
see ‘mind the gap’
No code as docs!
It is lazy and shows a lack of care to non technical users
“just read the code”
read – empathy for idiots
https://bit.ly/badcamp-write-the-docs

PIVOTING IN A PROJECT: STRATEGIES FOR ADJUSTING TO SCOPE CHANGES
BRIAN MANNING
KELLY DASSING

Over the last year or so of my life I have had the opportunity to lead a few projects as the Project Manager. It is always a fun time and I will take all the help I can get. This talk provided a very, very solid foundation for anyone hoping to get up to speed on how to manage large scale projects with multiple stakeholders. But all those lessons also work for teams of two and smaller deliverables as well.

Raw Notes:
Tips and tricks talk
Project team
architect, devops leads
creative director
digital strategist
accessibility lead
executive sponsor in the middle with a project manager to guide
kicking off
capturing the vision
help your client organize their ideas and get a plan of attack
make sure budget makes sense
realistic goals
all stakeholders agree,
failure is a domino effect a lot of the time
know what success looks like
ask the right questions early
kickoff agenda
who is responsible for each task
establish timeline and dates
what is a launch blocker vs nice to have?
who’s the captain?
Communicate with the whole team
establish what the project as a whole is even doing
make sure each role is represented at the table
Avoid the dreaded overcommitment
establish your MVP up front
know your deliverables
communicate with your product owner
don’t be afraid to say no
keep your team grounded in the delivery
don’t get caught with too many meetings
-maybe slack is better or an email for this?
managing team morale and motivation
don’t be afraid to say no
you need to push back to keep it on track
Hedwig helper’s hotline
everyday at 6:00pm EDT they would kick Kelly offline and someone steps in
keeping online all the time is bad, burn out
Identifying the goal and plan to accomplish it
revisit the MVP and adjust as needed
creating a Phase 2 plan shouldn’t be seen as a failure
show progress to stay aligned, established deliverables
refine roes and processes
keep your team on task with tools
Wrike, Trello, jira, etc
discovery, design, dev, QA, Deployment, post launch support
Agile Scrum framework at a glance
product backlog
sprint planning meeting
sprint backlog
1-4 weeks sprints
Sprint approval docs
little changes are inevitable, having docs makes this easier on everyone
continue clear communication to clients
take an extra 5 minutes to update tickets and communicate status
use weekly status emails to update project status
establish quick check-ins via a call or email to stay on target
getting ready to deliver success
check your deliverables against the established MVPs
Does the delivery match up with the success outlined at the start?
What does failure look like?
Are there elements of that that failure loks like present?
What happens if those pivots extend your time frame?
throw a wake, not a funeral
key to learn and grow on what was successful and what wasn’t
pivots cause you to try things that are outside your norm
did they work?
Multiple pivots can cause you to think on your feet and come up with new ways of doing things
why wasn’t the normal path working?
Postmortem can indicate failure even when successful

HOT DOG/NOT HOT DOG: ARTIFICIAL INTELLIGENCE WITH DRUPAL
Rob Loach

I feel that very few talks I go to are named this well. AI is the here and now of how things work so ‘automagically’ in so many aspects of our lives. The available tools are getting easier to use all the time while also getting more powerful. The ‘correctness’ scores of some of the images Rob tested were astounding. I love that we are getting to a point where the philosophy of “let the machines do the work” is applied to such specialized qualitative things.

Raw Notes:
Kalamuna
very distributed company
We keep teaching computers how to act as if they were human
Machine learning
recognize patterns
statistical analysis in a neural network
if at first you don’t succeed, try a billion more times
realtor.com – take pic of sign, not QR code
Patch.com – filter for obscene pic uploads
facial recognition – google recognized dogs, not just faces
Google Vision API
label detection to see objects
face detection
optical character recognition
explicit content detection
landmark detection
API gives you a breakdown on images, properties, safe search
pocks up recent changes well
stays current
and a whole lot of metadata
coordinates, and so on
score for these, confidence score
cloud.google.com/vision/
hot dog/not a hot dog app
Alan Turing is not a hotdog
a cucumber is not a hotdog
high confidence it is not
hot dog is a hot dog
93% confidence
kalamuna/hotdog – must set your own credentials with Google
Drupal module – sponsored by Pantheon
recipe website
used Umami
statue of david caught by adult filter
suspicious images can be approved
kalamuna/google_cloud_vision
code example looking at API

GITLAB HAS COME TO DRUPAL.ORG
Michael Hess

After some speculation if Michael would make it to camp, he arrived to great fanfare and put on his community hat, as opposed to his security hat to give us the state of the Drupal.org world as it moves from self hosted git servers to Gitlab for all the advantages it will bring. Are allt he wrinkles ironed out? It is being worked on and this move puts us on the right track for the growth of Drupal long into the future. If nothing else it is going to reduce the time it takes to go from newbie to committer. I am looking forward to keeping up with the news as this work evolves.

Raw Notes:
Better title, Gitlab is coming to D.o
little demo by end
some history
2001, Sunflower theme
CVS for code tracking
2005 – inline text in page and RCS, diffs
2008 – still in CVS, colse to present day patch
first drupalgeddon issue introduced
in 2011 – own git servers for Drupal
very complex project
48 hour outage to do the work
2018 – patch files still moving around
April 2008, Github went live
became standard
but why patches in the first place?
Easy and actually mailing list
2018 – for past 4 years been talking about how to get away from the patch system
Drupal issue queue is actually better than standard Github
issue queue makes it possible to jump in quickly compared to other projects
contribution credit
issues?
run own git servers, nothing really wrong but git servers hae never been kept up, security patched but architecture is old
code review
What are the goals
adopt a dev workflow that would be easy for folks outside Drupal
preserve unique elements
find a partner to help
wrote more requirements
workspaces model
again building own solution, own island again
let’s get off the island
why not just all over to Github?
works for some but not all, not many to one
leftPad
Looked at Bitbucket too
good but need add ons like Jira and Bamboo
but they were going to pick it
gitlab CEO reached out and wanted to help
formalized path forward and slack channels created
3 phases
Phase 1 replace Github services with Gitlab’s
if project maintainer, must change remotes
use SSH
drupal.org/gitauth
sandbox on d.o, workflow will need to change
Need to get HA worked out, Sync git history back to D.o
testing
push hooks made
phase 3
GitlabCI
confidential issues for security
GitLab issues
MatterMost (slack competitor)
project management boards
Phase 2 – merge/pull requests on Drupal.org
don’t work for DA but hoping end of January this is solved
built in IDE for Github
primary home for the Drupal project is still D.o
code, test results, git web viewer, merge requests = Gitlab
everything else = D.o
do we still need sandboxes?
how do we sync comments to just the relevant places?
still working on it
What are we forgetting? Something for sure

My session

LET’S LEARN GIT. NO MORE EXCUSES.

I went over time. I never do that typically, but I got excited during this talk. People asked very good questions and I see yet again how to improve on this presentation. It was thrilling to have people say they have not been able to grasp some of the concepts before on their own and nobody had just explained the whole big concept at once before. I am very proud of what this talk and how many people I have helped get over that line. It was the end of the camp I was the last session. Very grateful to the 20 or so people who stuck around to learn with me!

Wrapping up

BADCamp is a blur every year. There is a lot of build up and prep we do for it. Then once the event starts it is just an endless string of hello hugs, catching up for a few minutes and never feeling like you got that much done because you were so busy. From the opening setup to the last goodbye, it is a wonderful experience with the Drupal community really coming together to celebrate what we are and how we keep advancing. Drupal itself continues to evolve and so does the community, but I hope we never outgrow this sense of fun and folly that underlines events where we discuss serious concerns around the future. No matter what happens next for me, if at all humanly possible I will be back for BADCamp 2019!

Leave a Reply

Your email address will not be published.