Convert documentation to GAPDoc/AutoDoc, fix some errors in it - #32
Conversation
Replace the plain-TeX manual with GAPDoc XML built via AutoDoc, and drop doc/make_doc for a makedoc.g in the package root. See gap-system/gap#639. Converted with dev/gapmacro2gapdoc.py from the GAP repository, which corrected AlgebraByTable from a function to an operation and rewrote 31 TeX accents in the bibliography as Unicode. The rest was the examples, which the old build extracted to a file nothing ran: * A rec() spread over six lines had no "> " continuation prompts, so the extracted test read the first line as a whole statement, failed on the syntax, and took the remainder for expected output. * Four examples print a record containing a "time" field in milliseconds, and the walkthrough of the groups from [GLM24] ends in a break loop after a computation that, as the manual says, "was not progressing for several hours" -- with a path on the author's machine in the traceback. Those are <Log>. So are the examples that follow the pasted definitions of G and H, since the definitions are a <Listing> and never run. * KuroshAlgebra's example elides its progress output as "... some printout ..", which cannot pass as a test; kept as written, as <Log>. * The recorded output of the remaining examples has drifted. Refreshed from what GAP prints. Note that BinsByGTAllFields(2,6) now gives ten bins where the manual had eight: the shipping code says so, but it is a mathematical result rather than formatting, and worth a check. The test suite passes. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Also correct two formulas: |G| <= p^(p+1) had lost its bars, and C_G(G/Phi(G')) should be C_G(G'/Phi(G')) as elsewhere in the chapter.
The conversion put function names, calls, record components and globals into <M>, which renders them as italic maths. Documented functions now use <Ref>, the rest <C>, false <K>, and MIPBinSplit's parameters <A>. Also repairs constructs the conversion split across two <M> elements, such as <M>SmallGroup(p</M>^<M>n, i)</M>, and a stray MIPBinsSplit.
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #32 +/- ##
==========================================
+ Coverage 94.08% 94.09% +0.01%
==========================================
Files 62 62
Lines 26569 26569
==========================================
+ Hits 24997 25000 +3
+ Misses 1572 1569 -3 🚀 New features to boost your workflow:
|
olexandr-konovalov
left a comment
There was a problem hiding this comment.
To me this looks good. I have reviewed the process rather than the manual though - instead of comparing TeX and XML files, or comparing old and new PDF files, I have compared doc/autiso.tex and doc/autiso.xml which GitHub by some reasons nicely shows, and was content with that. I have scrolled through all diffs shown under this PR and looked at the resulting PDF in GitHub actions. All looks good, and trust that @fingolfin's conversion script is well tested.
|
@olexandr-konovalov, @fingolfin I would prefer to have it merged in "steps" so we can follow what model changed. |
|
@limakzi it's not a model, it is |
Resolves #9