Dave Karow

1.5K posts

Dave Karow banner
Dave Karow

Dave Karow

@davekarow

Build software faster without breaking things. Test what works (user behavior vs your opinion). Advocate for Feature Flags and Experimentation @SplitSoftware

San Mateo County, CA Katılım Ekim 2008
501 Takip Edilen468 Takipçiler
Dave Karow
Dave Karow@davekarow·
If you are in Chicago or can travel there, check out @GOTOchgo May 22-24. @GOTOcon shows are well run by a warm and hospitable team & always feature top-notch technical speakers on various topics. Here's a query for videos of the talks I've given at GOTO: google.com/search?q=youtu…
Dave Karow tweet media
English
0
0
1
89
Charity Majors
Charity Majors@mipsytipsy·
And of course there's @monkchips, coiner and popularizer of progressive delivery. Who else deserves credit when it comes to modern release practices? I'll make sure and link their names to the piece on Monday. Apologies for not asking beforehand. ☺️✌️
English
4
0
7
6.8K
Charity Majors
Charity Majors@mipsytipsy·
Hey, I am hearing through the grapevine that I should have cited or credited some of the folks who did a lot of work in this area, and this is super true!! It looks like we have @LaunchDarkly to thank for "separate deploys from releases", and I know @wiredferret did a ton there.
Charity Majors@mipsytipsy

You may have seen me rant about other deploy-induced atrocities, but this one deserves special emphasis. 🌺 Deploys 🌸 Are 🌼 The 🌹 WRONG 🌷 WAY to release new features, toggle configs, or otherwise make changes that impact user experience. honeycomb.io/blog/deploys-w…

English
4
4
55
31.2K
Dave Karow
Dave Karow@davekarow·
@jeremydavis @housecor The Rollout Board in @SplitSoftware calls them "Ready to Cleanup" and has a built-in query that calls attention to them. Still, having them march slowly toward you moaning "brains!" could have it's appeal too 🤷‍♂️
English
1
1
3
171
Cory House
Cory House@housecor·
Problem: Feature flags are handy, but it’s easy to forget to remove them later. This leads to needless overhead, complexity, and risk. Solution: Write an automated test that fails if the flag still exists after a given date.
English
72
119
1.3K
186.2K
Dave Karow
Dave Karow@davekarow·
Hey ⁦@afitnerd⁩ check out the t-shirt ⁦@davefarley77⁩ is wearing in this interview about the value of real #ContinousDelivery. Fellow PacMan fan it seems! Love the framework you laid so I could do my How PacMan Taught Me To Love Progressive Delivery talk a while back.
Dave Karow tweet media
English
1
1
6
2.2K
Jeffrey Chupp
Jeffrey Chupp@semanticart·
I'm convinced Feature Flags are one of the best tools you have to mitigate risk. Roll something out to 1% of users and your bug tracker and alerting goes wild? Guess what? 99% of your users never had an issue. 😎 You should be using Feature Flags more.
English
27
42
460
77K
Dave Karow
Dave Karow@davekarow·
@Codydearkland @LaunchDarkly Hey Cody, what you describe in the first half of your video is actually Blue/Green, not A/B. You can't conduct an A/B test by sending 100% of traffic to B. Without a split of traffic (50/50 or otherwise), there is no way to compare A and B performance. Check out @ronnyk for more.
English
2
0
2
48
Dave Karow
Dave Karow@davekarow·
@sesteva @patricioe @ph1 @SplitSoftware 2/2: For example here is that section from JavaScript SDK docs: #using-the-sdk" target="_blank" rel="nofollow noopener">help.split.io/hc/en-us/artic… - Using the SDK --Basic use --Shutdown --Multiple evaluations at once --Attribute syntax --Binding attributes to the client --Get Treatments with Configurations
English
1
0
1
28
Dave Karow
Dave Karow@davekarow·
@ph1 @patricioe @SplitSoftware Those who haven’t gone that route try to shift the conversation to whether the evaluation rules are stored locally. That’s a red herring. Rules or not, everyone’s local app code must contain “if flag evaluates to x, do y” so the list of x’s and y’s is there for smarties to find.
English
0
0
1
44
Dave Karow
Dave Karow@davekarow·
@ph1 @patricioe @SplitSoftware Security is a *fun* topic. Split.io is architected to never send PII out of a client SDK to the cloud for evaluation, which means we never see it, let alone decide whether to store it.
English
1
2
2
146
Dave Karow
Dave Karow@davekarow·
@ph1 @patricioe @SplitSoftware When evaluation is done at the very moment the user reaches the flagged code, records of *that* evaluation are meaningful. That’s not the case with evaluations done and cached at launch.
English
0
0
0
17
Dave Karow
Dave Karow@davekarow·
@RaulJuncoV @SplitSoftware Finally, since you'll find a lot of talk about using data to make sense of what's happening with your flags, this security primer that shows how we pull that off without storing PII on our end might come in handy. split.io/wp-content/upl…
English
1
0
2
15
Raul Junco
Raul Junco@RaulJuncoV·
90% of developers think that Feature Flags are just for hiding functionality, and they are wrong. Feature Flags is a powerful solution to decouple your Deployments from your Releases. 5 facts about Feature Flags you need to know to take full advantage of them.
English
20
23
198
73.1K