fbyrne.bsky.social
38 posts
31 followers
105 following
Regular Contributor
Active Commenter
comment in response to
post
I saw this post and immediately picked the book up. I have lived in the presence of others who would constantly chirp “hope is not a strategy” and not go further.
It honestly feels like leaving Plato’s cave and I’ve only read one chapter. Thank you!
comment in response to
post
time tariffs
comment in response to
post
:| what happened to all the cool conferences.
comment in response to
post
Dark portents are gathering.
comment in response to
post
3. groupby aggregates return dataframes (cont)
Maybe I'm just a fan of Monads like Maybe or Option where you specifically have to unwrap the type once you've entered the Monad's universe.
I think I'm missing something here because I see polars follows pandas here. I think Julia also would follow.
comment in response to
post
3. groupby aggregates return dataframes
This one is really irritating as a DataFrame diff has no concept of any groupings: it just diffs each row. So I was calling diff for several hours and not understanding why I had "incorrect" results.
So now I know to do df.groupby().sum().groupby().diff()
comment in response to
post
2. groupby criteria ordering matters
["Project", "Year", "Month"] is very different from ["Year", "Month", "Project"]
comment in response to
post
1. groupby using Datetime (continued)
Personally I would expect the groupby to automatically maintain type consistency by default. After all, Datetime objects do have an ordering, right? Why convert it to a string? Now I get surprised and have to do more work.
comment in response to
post
1. groupby using Datetime
I was a bit dumbfounded when I grouped by a datetime Column only to have the new MultiIndex be a string. You specifically have to use the datetime accessor to pull out individual components of the datetime to use.
i.e. groupby(df["Date"].dt.month)
comment in response to
post
Here's an overview of what I learned:
1. How to groupby using Datetime. (is the API good here?)
2. groupby grouping criteria ordering matters. (skill issue)
3. groupby aggregate functions return a dataframe. (is the API good here?)
comment in response to
post
I wonder if you could make a model that understands gzipped source files.
comment in response to
post
There it is. Each department has their own practices.
handbook.gitlab.com/handbook/eng...
comment in response to
post
Prioritization:
At the end of chapter 4 backlog prioritization is just that important things move to the top and items > 6 months old get dropped.
Love it.
comment in response to
post
Deliver Often:
Key to my heart these days. Though we need to be more specific. These are small incremental changes ideally delivered to the customer.
David uses the example of texting your date right after (better value) to show interest versus a larger display with a greater delay (less value).
comment in response to
post
Note: David’s prescription is to allocate time for collaborative design each day. He also references Scott Ambers Agile Modeling.
TODO: read agilemodeling.com
comment in response to
post
Focus on quality:
One key area for improvement I see for myself is encouraging collaborative design and analysis.
I feel it is common to just DIVE into writing code. Is it because of pressure? Lack of guidelines facilitating design?
TODO: survey teams on the state of collaborative design.
comment in response to
post
David’s “recipe for success” has given me some items to reflect on:
1. Focus on Quality
2. Reduce work in progress
3. Deliver often
4. Balance Demand against throughput
5. Prioritize
6. Attack Sources of Variability to improve predictability.
comment in response to
post
Can we talk about how much cigarettes are showing up in the media these days as well?
comment in response to
post
But then how do you associate the deployment with a jira ticket?
Fortunately the Jira team is pretty open with this. It requires iterating all commits between now and the previous deployment.
So I’m imagining custom labels injected to mark commits.
github.com/atlassian/gi...
comment in response to
post
Now these are interesting. Connecting basic tasks to updates in jira.
I especially like putting a ticket on hold if a deploy fails.
www.atlassian.com/software/jir...
comment in response to
post
This is similar community.atlassian.com/forums/Jira-...
comment in response to
post
I like the companion long form. I’m curious about shying away from a checklist. Certainly true it can’t guarantee a promotion, but suggested activities have felt appealing to me.
comment in response to
post
This is why SCIM is an enterprise feature.
comment in response to
post
This is fantastic. I’ve always been interested in organizing a smaller conf. I may borrow your model :)
comment in response to
post
I would love to buy a xiaomi they are way cool.