[us-commits] [ehb54/ultrascan3] ab5571: extrap_c0: encode every result-changing option in ...

emre brookes noreply at github.com
Tue Jul 21 14:32:01 MDT 2026


  Branch: refs/heads/ehb54-issue-920
  Home:   https://github.com/ehb54/ultrascan3
  Commit: ab5571e2f6ed60a083b1f7f2d2076acf0260bddb
      https://github.com/ehb54/ultrascan3/commit/ab5571e2f6ed60a083b1f7f2d2076acf0260bddb
  Author: ehb54 <brookes at uthscsa.edu>
  Date:   2026-07-21 (Tue, 21 Jul 2026)

  Changed paths:
    M us_somo/develop/src/us_hydrodyn_saxs_iqq_extrap_c0.cpp
    M us_somo/somo/doc/manual/somo/somo_saxs_extrap_c0.html

  Log Message:
  -----------
  extrap_c0: encode every result-changing option in the output curve name

Reported from testing: a GCV run and a no-GCV run produced the SAME curve name, so the
two were distinguishable only by an arbitrary -1 suffix. Confirmed in our own regression
set, where recip.dat and recip_nogcv.dat both carried "Sub_extrap_c0_recip".

Auditing the namer against the option list found four options that change the result but
were absent from the name:

  _nogcv          automatic slope regularization switched off
  _fb<N>          manual fit-broadening window (only active when GCV is off)
  _refc<c>        an explicitly chosen reference curve, tagged with its concentration
  _rawres         the outlier leverage correction disabled -- appended to _qcN because it
                  can change WHICH curve was dropped, so two runs could otherwise both
                  read _qc1 while resting on different curves

Tokens are emitted only for non-default choices, so an ordinary run keeps a short name.
The comment on the namer now states the invariant explicitly -- every option that can
change the result must appear here -- since the failure mode is silent and only shows up
later as two curves that cannot be told apart.

Verified: all 9 regression combos now yield unique names (previously two collided), and
the new tokens appear correctly (_refc1_8179, _nogcv_fb9, _qc1_rawres vs _qc1). Exactly
one baseline changed, name-only with 0 data-line differences, and was refreshed. Manual
updated; all 15 tokens the namer can emit are now documented.

Co-Authored-By: Claude Fable 5 <noreply at anthropic.com>



To unsubscribe from these emails, change your notification settings at https://github.com/ehb54/ultrascan3/settings/notifications


More information about the us-commits mailing list