Pull requests versus pair programming: Difference between revisions
From YM2149.org
Jump to navigationJump to search
(One intermediate revision by the same user not shown) | |||
Line 1: | Line 1: | ||
== Pull requests == | == Pull requests == | ||
* | * It's maybe possible for [[pull request]]s to be a cheaper way of achieving the same benefits as pair programming | ||
* But this requires a lot of discipline from the developer and the reviewer | * But this requires a lot of discipline from the developer and the reviewer | ||
* | * Nitpicks are like [https://www.simpsonsarchive.com/episodes/2F11.html 30 million dollars for the perverted arts] | ||
== Pair programming == | == Pair programming == | ||
* | * By far the best way to transfer knowledge between senior and junior | ||
** | ** And it goes both ways e.g. junior will share some sick modern tricks | ||
* | * Keyboard shortcuts, cli tricks and so on stick in the mind much more readily when learned by doing with guidance | ||
** | ** Similarly, a knowledge transfer session is much more effective when the audience does the steps guided by the presenter | ||
* Pair programming lifts everyone to the best performer and beyond, pull requests drag everyone down to the worst performer | |||
== Good article == | == Good article == | ||
* https://mtlynch.io/code-review-love/ of which | * https://mtlynch.io/code-review-love/ of which I 100% agree with most points | ||
* | * Writing a description should be a simple case of linking back to the ticket | ||
* | * Put cosmetic changes (if absolutely necessary) in their own commit so reviewers don't have to consider them | ||
* | * The reviewer may have fresh eyes, but none of the context, and should appreciate the author has made decisions based on a much more thorough understanding of the situation | ||
[[Category:Programming]] | [[Category:Programming]] |
Latest revision as of 20:26, 22 February 2025
Pull requests
- It's maybe possible for pull requests to be a cheaper way of achieving the same benefits as pair programming
- But this requires a lot of discipline from the developer and the reviewer
- Nitpicks are like 30 million dollars for the perverted arts
Pair programming
- By far the best way to transfer knowledge between senior and junior
- And it goes both ways e.g. junior will share some sick modern tricks
- Keyboard shortcuts, cli tricks and so on stick in the mind much more readily when learned by doing with guidance
- Similarly, a knowledge transfer session is much more effective when the audience does the steps guided by the presenter
- Pair programming lifts everyone to the best performer and beyond, pull requests drag everyone down to the worst performer
Good article
- https://mtlynch.io/code-review-love/ of which I 100% agree with most points
- Writing a description should be a simple case of linking back to the ticket
- Put cosmetic changes (if absolutely necessary) in their own commit so reviewers don't have to consider them
- The reviewer may have fresh eyes, but none of the context, and should appreciate the author has made decisions based on a much more thorough understanding of the situation