Repository navigation
To test
andy edited this page Oct 24, 2026
·
103 revisions
Last updated: 2026-10-24 afternoon. Hand checks waiting for a person at the screen. An item is added when a fix is merged (the issue is closed by the fix commit) and removed when it has been checked; anything that fails goes back to its issue. The build under test is named in each section.
How to help: anyone can take an item. Build or install the version named in the section (packages from the
latest CI run on main, or build as in GETTING_STARTED.md),
try the check, and comment on the linked issue: what you did, what you saw, operating system and version, and a
screenshot if something looks wrong. Items are taken off this page once someone has reported them as working.
| Check | What to look for | Issue |
|---|---|---|
| Copy numbers | Organizer and Calculation Summary values (e.g. final energy) can be selected with the mouse and copied; table cells open a read-only field on double-click | #408 |
| Constraints in the Calculation Editor | Geometry constraints are listed for every run type; a Geometry run with scan constraints says it is a relaxed scan | #456, #461 |
| TS from two calculations | Show mapping works without a Save first; a warning when reactant/product were optimised at another level | #464, #468 |
| ECCE-QM TS search | TS search from one structure, frequencies after, one imaginary mode | #452 |
| UV-vis | TD-DFT options in the ORCA and Gaussian 16 editors; spectrum plotted, transitions labelled HOMO/LUMO | #383 |
| GROMACS groups | Rerun Dynamics with Centre of mass removal groups = Defined groups and no groups, Save: the .mdp starts with the "; Note: ... System is used" line | #482 |
| QM step plots | Geometry Trace: energy relative to step 1 in kJ/mol, final energy in the title, whole-number steps; Step plots: Energy change on log scale; zoom out with the wheel stops at the data, pan stays inside | #495 |
| Gaussian trace, one frame per step | Reconnect a G16 optimisation: number of trace points = number of steps; clicking point n shows step n | #498 |
| Allocation in the Machine Browser | Machine settings > Kebnekaise > Connection: Allocation command = projinfo, Save; Machine Browser > Kebnekaise > Allocation shows projinfo's output | #494 |
| Duplicate with Final Orbitals, NWChem | Run an NWChem SCF or DFT job, then Duplicate with Final Orbitals: the copy reads guess.movecsin and converges in a few iterations; with another basis set the Calculation Editor warns | #407 |
| Failed submit reason | A failed submit (e.g. wrong run directory) leaves the reason in the Organizer status line of the Ready calculation; a good relaunch clears it | #516 |
| MD averages | GROMACS dynamics: Progress box and summary show average ± SD (minimisation: last value) | #517 |
| Dynamics tab | Ensemble row and hint on their own lines, no overlap with Data Collection Steps | #515 |
| Check | What to look for | Issue |
|---|---|---|
Polyoxometalates library (branch wip/405-pom) |
Builder → Structure Library → Polyoxometalates: each entry opens, looks like the named ion, bonds sensible; the list of missing species is on the issue | #405 |
| Check | What to look for | Issue |
|---|---|---|
| Vibrations spectrum (Windows 11; maintainer test bed, not for the Linux deb) | Spectrum shows one curve; peaks can be picked at once | #445 |
| Check | What to look for | Issue |
|---|---|---|
| Gaussian 16 Cartesian optimisation | Theory 'Use Cartesian Coordinates' gives Opt=(Cartesian) in the route (seen in the input) and the job converges |
#488 |
| Check | What to look for | Issue |
|---|---|---|
| Install from scratch | RHEL 9, Ubuntu 24.04, Debian trixie: client, server, each deployment mode | #193 |