#199 moved both images to the Anaconda 2026.07 baseline. The docs still quote the 2026.06 sizes, which containers/README.md now dates as measured before the move. The image-size job in test-container.yml measured the new images on 2026-09-25 (run 36122900672):
| Image |
Compressed pull |
On disk |
quantecon (full) |
3.39 GB (was 3.33) |
8.79 GB (was 8.60) |
quantecon-build (lean) |
2.85 GB (was 2.93) |
7.10 GB (was 7.32) |
The lean image is now about 16% smaller to pull than the full one, not the ~12% containers/README.md:12 says.
The figures are hard-coded in 22 lines across 10 files, so every baseline or TeX Live change leaves most of them stale:
| File |
Lines |
README.md |
153, 154 |
containers/README.md |
9, 10, 12 |
containers/quantecon-build/README.md |
18, 79, 80 |
docs/CONTAINER-GUIDE.md |
9, 10, 64 |
docs/QUICK-REFERENCE.md |
23, 24 |
docs/dev/ARCHITECTURE.md |
15, 54 |
docs/dev/CONTAINER-VALIDATION.md |
17, 18 |
docs/dev/TESTING.md |
187 |
setup-environment/README.md |
34, 35 |
.github/copilot-instructions.md |
33, 34, 135 |
Deferred here from Copilot's review of #199; the replies on that PR point at this follow-up.
Assisted-by: Claude Code (Claude Opus 5.5)
#199 moved both images to the Anaconda 2026.07 baseline. The docs still quote the 2026.06 sizes, which
containers/README.mdnow dates as measured before the move. The image-size job intest-container.ymlmeasured the new images on 2026-09-25 (run 36122900672):quantecon(full)quantecon-build(lean)The lean image is now about 16% smaller to pull than the full one, not the ~12%
containers/README.md:12says.The figures are hard-coded in 22 lines across 10 files, so every baseline or TeX Live change leaves most of them stale:
README.mdcontainers/README.mdcontainers/quantecon-build/README.mddocs/CONTAINER-GUIDE.mddocs/QUICK-REFERENCE.mddocs/dev/ARCHITECTURE.mddocs/dev/CONTAINER-VALIDATION.mddocs/dev/TESTING.mdsetup-environment/README.md.github/copilot-instructions.mdcontainers/README.mdalready states the method and the date) that the other docs link to, or refresh every mention in place.Deferred here from Copilot's review of #199; the replies on that PR point at this follow-up.
Assisted-by: Claude Code (Claude Opus 5.5)