Erratum
Paolo Ferragina, Francesco Tosoni. The Energy–Throughput Trade-off in Lossless-Compressed Source Code Storage. SANER Companion 2026: 157–164.
This page documents corrections to the published version. Fixes are graded by severity, following the convention popularised by Gonzalo Navarro:
- small technically incorrect statements that do not affect the results at all.
- medium issues with some impact, but the main results stay correct.
- serious issues that compromise the main results.
Both items below are small and self-contained: they concern axis labels and one table column only, and leave the experiments, discussion and conclusions of the paper unchanged. Both have already been corrected in the extended version of this work.
Energy axes off by a factor of 100 in Figures 3 and 4
Figure 3 (green subplots) · Figure 4 (right-hand axes)
The energy axes of the green subplots in Figure 3 and the right-hand axes in Figure 4 were mislabelled by a factor of 100. Only the numeric scale of those axes is affected: the plotted curves, their shapes and every relative comparison drawn in the text remain valid.
Cause: a regular expression that did not account for the decimal separator (two digits after the point) in the energy readings, so it extracted values 100× too large.
Doubled insertion-throughput column in Table II
Table II (rightmost column)
The rightmost column of Table II reported insertion-throughput values that were twice their true magnitude. The correct figures were already plotted in Figure 5, which is authoritative.
Cause: most likely a mis-scoped grep over the terminal logs that counted the inserted records twice.