I have not been following "the book."
- I have not been maintaining the build documentation well after I setup Jenkins.
- I am still relying on an old shell script for configuration.
- I have maintained code that builds using Microsoft Visual Something Community Edition instead of a more sane alternative for runtime performance reasons.
There are many other things that I could list too.
But perhaps worst of all:
- I merged a PR that was not yet upstreamed, told people I would upstream it, and haven't gotten around to doing so.
Seriously, upstreaming first is everyone's responsibility. Even if it is just a code-cleanup patch to remove support for 1990's hardware (CRAY.)
If a fortune 500 company can accept an upto 50% "performance" reduction towards new feature development by upstreaming 1st, then I can too in my free time.
We are making great progress towards upstreaming nx-libs into X.org, largely by unbundling both modified and unmodified dependencies.
All told, developing a feature this way may take about twice the time and effort it would just to develop it in private, off a stable release, on a vendor branch.
86MB uncompressed -> 32MB uncompressed :)
However, code cleanup that we make to our (inherited) fork of X.org should be done upstream 1st. We have actually increased our delta against upstream by not doing so.
There are numerous other things I have been doing wrong that I learned at Flock. I am pondering far greater changes to my workflow.
SPARROW: You have taken the first step on a path back to righteousness. In light of this ...EDIT: That wasn't a good link for CRAY. So I just specified "CRAY" in text.
No comments:
Post a Comment