Commits for braney

ba2a019284cf8b41b243f3a8817763c23d870965 Mon Jun 15 14:54:55 2026 -0700

prevent hgTracks crash when rearrangement and density graph are both enabled, refs #36668 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- src/hg/hgTracks/snake.c - lines changed 6, context: html, text, full: html, text

36fbb0bc581f5189dddc2c2a68aea11a378ad8fc Wed Jun 17 06:54:40 2026 -0700

prevent docker image build failure when one mirror is down by adding curl timeouts and || true to the genome vs genome-euro latency check in browserSetup.sh, refs #37780 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- src/product/installer/browserSetup.sh - lines changed 4, context: html, text, full: html, text

7bed537d83b701bb4284080c2b634f1069417cae Thu Jun 18 12:07:03 2026 -0700

fix squishyPack track stealing a neighboring track's center label when image orders collide A squishyPack track is split into a full-height part and a squished part that are stitched back together by smashSquish(), which assumes the squished clone is drawn immediately after its source. flatTracksCmp() broke an image-order tie involving a squink track purely by visibility, even against an unrelated track sharing the same order, so a colliding track (e.g. MANE vs GENCODE V49) could sort between the source and its squink. smashSquish() then merged the squished slice onto the wrong row, dragging the neighbor's center label with it. Keep a squink adjacent to its own source in flatTracksCmp(), and guard smashSquish() so it only merges a linked neighbor cloned from that track, refs #37785
- src/hg/hgTracks/imageV2.c - lines changed 15, context: html, text, full: html, text

3e922f07e4d8c51a3b989e3dc6748e33f8516e3b Mon Jun 22 09:10:05 2026 -0700

Document two -O3 cleanup details raised in v500 code review Follow-up to code review #37767, addressing two points on commit f57fc11d951: - The else guards in sanger22gtf.c and bottleneck.c looked like leftovers from before errAbort was marked noreturn, but they are still required: removing either one reintroduces -Werror=format-overflow at -O3, because that warning runs before the noreturn-based dead-path pruning. Added a comment at each else saying so. - makeGrayShades (hgGene/altSplice.c and hgTracks/simpleTracks.c) writes shadesOfGray[maxShade+1] = MG_RED as an overflow sentinel. Spelled this out in both function descriptions so the maxShade+2 array sizing is documented at the function, not just at the caller. Comment-only; no behavior change. refs #37767 refs #37761
- src/hg/hgGene/altSplice.c - lines changed 3, context: html, text, full: html, text
- src/hg/hgTracks/simpleTracks.c - lines changed 3, context: html, text, full: html, text
- src/hg/makeDb/outside/sanger22gtf/sanger22gtf.c - lines changed 1, context: html, text, full: html, text
- src/weblet/bottleneck/bottleneck.c - lines changed 1, context: html, text, full: html, text

0939cd926f6ff4d587eafff011363540d2b05b53 Mon Jun 22 10:50:23 2026 -0700

Add -g to the default COPT so production -O3 builds keep debug symbols Default COPT becomes "-O3 -g". -g only emits DWARF debug sections; it does not change code generation, so optimization is unaffected and the binary keeps gdb symbols and meaningful apache stack traces. An explicit COPT on the command line or in the environment still overrides it. Validated on hgwbeta by rendering a heavy Recommended Track Set session (Non_Coding_SNVs hg38) at two ranges with -O3 vs -O3 -g binaries, 12 interleaved iterations each: the two straddle each other within run-to-run noise (O3g 2.5% faster on one range, 0.9% slower on the other), i.e. no measurable slowdown. refs #37767 refs #37761
- src/inc/common.mk - lines changed 1, context: html, text, full: html, text

98572a57d651f8256c0d21e690f5df23fc1ce3c7 Tue Jun 23 16:05:31 2026 -0700

hgTracks: show lifted (target) coordinates in quickLift bigBed mouseOvers, refs #37615 Under quickLift the bedRow used for mouseOver substitution is loaded from the source-assembly interval, so $chrom/${chromStart}/${chromEnd} reported the pre-lift position while the position bar showed the lifted position. Overwrite those three coordinate fields with the lifted bed's target coordinates before substituting, so the tooltip matches where the item is drawn. Replaces the earlier disclaimer note that flagged the discrepancy.
- src/hg/hgTracks/bigBedTrack.c - lines changed 40, context: html, text, full: html, text

48d45fa3cbdddcb62775f1fe01eef865407e035e Wed Jun 24 13:47:07 2026 -0700

36b9de7c24cba2dfabb7fccb49abf496c5835e5c Thu Jun 25 08:57:23 2026 -0700

Fix single-click on ruler track to zoom regardless of the highlight-dialog setting, refs #27113 A single click on the base position ruler updated the displayed position text but did not zoom when the Enable highlight with drag-and-select option was on (the default). selectEnd computed the zoom window and wrote it to the position display, then called selectionEndDialog, which returns early on a single click (startTime null), so nothing navigated. The dialog branch never checked for a ruler click even though its comment said it should. Add that check so a ruler click always falls through to the zoom path.
- src/hg/js/hgTracks.js - lines changed 1, context: html, text, full: html, text