Qiskit / qiskit-fermions / 34597204063
98%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 113
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 97.958%. Remained the same
34597204063

push

github

web-flow 
Categorize release notes (#345)

* docs: introduce release note categories

* docs: sort existing release notes into the new categories

The categories introduced in the previous commit split the Python and the C
API, so a note documenting both now needs one bullet per section rather than
one bullet mentioning both. Sections render far apart on the page, which is
what forces the rephrasing here: a bullet that read "the C API gained the
matching accessors" or "added the same three mappers to the C API" was only
legible next to the Python bullet it followed, and the trailing "In the C API,
..." paragraphs of the grouping notes had no home once split out. Each C bullet
now restates its own subject, and the C FCIDump note repeats the parse failures
and the ab-packing caveat it used to defer to the Python bullet for.

Two notes also move out of the API sections entirely. The exact FCI trace
changes no API and no semantics -- it only improves the conditioning of an
existing simulation path -- and dropping the by-value copy from the
Jordan-Wigner binding is likewise invisible except in what it costs, so both
belong under performance rather than features. The memory blowup they sat next
to stays a bug fix: mapping a large Hamiltonian could exhaust memory outright.

The Jordan-Wigner memory fix now names the C entry point too. The fix landed in
core, so `qf_ferm_op_jordan_wigner` inherits it, and the note read as though
only Python were affected; the same commit had already documented the behaviour
in the C docstring but filed no note for it. The shared fixes and performance
sections are the right home for it -- one core fix is one fix, and splitting it
would duplicate the paragraph -- so only the subject needed widening. The
by-value copy in the second bullet stays Python-only: the C API always passed a
pointer and never had that copy. While there, the caveat that the result is no
longer fully simplified moved into the note, since a caller who needs every
dup... (continued)

12088 of 12340 relevant lines covered (97.96%)

534165.72 hits per line

Jobs
ID Job ID Ran Files Coverage
1 34597204063.1 113
97.96
 GitHub Action Run
Source Files on build 34597204063