Add a sorted wordcount to the bottom of all_pod_files_spelling_ok

[kentfredric]

This makes nuking the most commonly misspelt words easier if you're
working on a large project with a lot of files, by helping you easily
see what needs to be deemed a "stopword" and what needs to be hit with
an edit pass.

This was an an apparent issue when I was adding a pod spelling check to dbix-class, which had 80 subtests worth of spelling failures which was hard to make sense of.

After this patch, it emits the following at the end:

# All wrong words: ActorRoles=1, Authorized=1, AutoCast=1, BelongsTo=1,
#    Bloggs=1, CachedKids=1, Centos=1, Compat=1, DBA=1, DELETEs=1, DEV=1,
#    Easysoft=1, Eg=1, FC=1, FileColumn=1, FirstSkip=1, FromForm=1, HRI=1,
#    INSERTs=1, Inflators=1, LimitXY=1, LiveObjectIndex=1, MRO=1,
#    ManyToMany=1, Microsft=1, Multicreate=1, MyApp=1, NULLs=1,
#    NoObjectIndex=1, Northwind=1, Optimizer=1, Overridability=1,
#    POSTGRESQL=1, Queryable=1, RDBMSes=1, RDMS=1, REPL=1, RESULTSET=1,
#    ReLoad=1, Reblessing=1, ResulSet=1, ResultSetColumn=1, Resultset=1,
#    Retrevial=1, RowNum=1, SQLT=1, SchemaVersions=1, Serializable=1,
#    Stringfy=1, Subquery=1, Subselects=1, Suported=1, TMTOWTDI=1,
#    TxnScopeGuard=1, UPDATEs=1, VARCHAR=1, WhereJoin=1, YYYY=1, albumid=1,
#    artifact=1, authorization=1, autodetected=1, autodetection=1,
#    autodetects=1, autoinc=1, autovalidation=1, blabla=1, bugreport=1,
#    caling=1, callframe=1, catalyze=1, cdbi=1, chocolateboy=1, columnname=1,
#    dbh=1, dclone=1, dcloned=1, dcloning=1, de=1, deallocating=1,
#    deferrable=1, dsn=1, explorative=1, failover=1, gR=1, generalized=1,
#    getter=1, gravatar=1, hundres=1, inflator=1, initialize=1, insertdb=1,
#    instantiation=1, introspectible=1, lifecycle=1, localize=1, logsystem=1,
#    materialized=1, mis=1, mysqld=1, nfreeze=1, nls=1, noindexed=1,
#    numification=1, onwards=1, optimized=1, optimizer=1, organize=1,
#    params=1, qw=1, randomizer=1, readbound=1, rebless=1, recognize=1,
#    reconnectable=1, reimplementation=1, ruleset=1, scalarref=1, se=1,
#    serializable=1, serialize=1, serialized=1, serializing=1, specialized=1,
#    sqlt=1, stacktrace=1, standardized=1, suboptimal=1, superset=1, sys=1,
#    sysdate=1, testdb=1, thusly=1, transactionally=1, txn=1, uncommited=1,
#    uninflated=1, uniqueidentifierstr=1, unixODBC=1, unserializable=1,
#    upto=1, utilize=1, versioning=1, webpages=1, xyz=1, AutoCommit=2,
#    BUILDARGS=2, DBs=2, DESC=2, EasySoft=2, FK=2, FilterColumn=2,
#    HashRefInflator=2, Kioku=2, LimitOffset=2, LinerNotes=2, ORMs=2,
#    PREFETCHING=2, Postgres=2, RHS=2, ResultSets=2, SELECTs=2, SkipFirst=2,
#    Subqueries=2, attr=2, autoincremented=2, autoincrementing=2,
#    balancers=2, callsites=2, colname=2, ddl=2, debugcb=2, fastpath=2,
#    freeform=2, incrementing=2, initialized=2, nextval=2, normalize=2,
#    nullable=2, overwritable=2, postgresql=2, prefetching=2, recognizes=2,
#    reconnection=2, resultsource=2, serialization=2, stringifiable=2,
#    syntaxes=2, uninserted=2, varchar=2, DBDs=3, Extensibility=3,
#    FetchFirst=3, InflateColumn=3, Informix=3, JOINs=3, PKs=3, Replicants=3,
#    RowNumberOver=3, TT=3, attrs=3, behaviors=3, natively=3, prefetches=3,
#    preversion=3, recognized=3, subquery=3, subref=3, subselects=3, unary=3,
#    uniqueidentifier=3, DSNs=4, GenericSubQ=4, RHEL=4, Replicant=4, cond=4,
#    debugfh=4, debugobj=4, overridable=4, reblessed=4, subqueries=4,
#    unversioned=4, ORM=5, ResultSources=5, callsite=5, normalizing=5,
#    classdata=6, eg=6, subselect=6, Firebird=7, prefetched=7, savepoints=8,
#    Balancer=9, DDL=9, LongReadLen=10, balancer=10, resultsets=11,
#    savepoint=13, sql=14, behavior=15, unicode=17, ODBC=23, replicants=42,
#    replicant=43, resultset=163

Which indicates resultset is either a good candidate for a stopword or a universal translation =).

Though "All wrong words" seems awkward and forceful, but I couldn't think of a better suggestion.

[dagolden]

Cool! I'd want to see this in descending order, though.

[kentfredric]

@xdg, Ascending order was chosen mostly because of how terminal scroll means the most significant are more likely to stay "in screen" at the end of a test run, though I'll gladly flip the orientation if its deemed necessary =)

I could probably make it nicer still by grouping by frequency instead of having a mass of =1.

[dagolden]

That's a good point. Most of the time I have only a dozen or so words so it doesn't really matter. So I take it back. I'm neutral on the direction.

[kentfredric]

This is the revised format I'm working on, I think its much easier on the eyes:

# All wrong words:
#      1: ACHTUNG, ASC, Analyzes, AoA, BCP, Backcompat, BegunWork, Bowden,
#         Caelum, ColumnCase, DBICTest, DBICs, DBIHacks, DDL, DESC, DTRT,
#         DWIW, De, FC, FIME, FIXUP, FKs, ForceUTF, GCed, GenSubQ, HELEMs,
#         IC, IFF, Informix, JOINs, LOBs, MOAR, MSAccess, MoreUtils, MsSQL,
#         NULLs, Normalize, ORed, ORing, OpenClient, PKs, POC, RDBMSes,
#         RSC, RaiseError, RestrictWithObject, ResultObject,
#         ResultSetColumn, RowNum, Runmode, STRICTMODE, Savepoints,
#         TEXTSIZE, TxnScopeGuard, WHEREs, WHOREIFFIC, al, aliastype,
#         aliastypes, anchore, artistid, asshats, bcp, bindattrs, bindlist,
#         bindtypes, bindvals, blockrunner, brainer, bugwards, bulkLogin,
#         captainL, centered, classdata, codepaths, codition, colinfos,
#         collapsable, collapsers, collist, confient, crosstable, curcuit,
#         dbi, defeaults, deferrable, deflator, deflators, depping,
#         dequalify, dir, disemvowel, eiehter, equalities, fBSD, fc,
#         firstcol, fixup, fsvo, ftds, fucktards, fugly, fulfill, groditi,
#         hotspot, hve, idcols, iff, implmentation, inambiguous, inflators,
#         initializations, interbase, introspectable, ivsize, jnap,
#         joinpath, joinstructure, jpath, keynames, lhs, libs, libsqlite,
#         localization, memleaks, millisecs, minimize, misbehavior, mkpath,
#         multicol, multiplicator, nonbind, normalized, nullable, objlike,
#         param, parameterize, params, parenthesized, parenthesizing,
#         pessimization, pks, pov, prefetches, preload, preloaded,
#         premulti, premultiplication, premultiplies, pruneable,
#         pseudoforking, qsub, rbuels, reblessed, recursing, reenabling,
#         refactor, refcount, refcounts, reinstall, reinvoke, reinvokes,
#         relname, relnames, relobjs, renderer, replicant, replicants,
#         reselect, reselected, rewriter, ro, rownum, rowparser, rset, rv,
#         sanitize, sanitized, scalarref, scrutinize, sensical,
#         serializing, somesuch, soooooo, stabilize, stackable,
#         stringifiable, subquerying, subselects, subst, sucky, swindon,
#         synchronize, synchronizing, tae, temporarly, tempvars, th,
#         toplevelness, unicode, unparseable, unqualify, unresolvable,
#         utilizing, vpp, vxs, waaaaay, wi, wo, wtf, yay, yyyzzz
#      2: CLOB, CursorType, DBA, FK, Firebird, HRI, JF, LobWriter, SEL, Stil,
#         adOpenStatic, analyze, autoinc, behavior, bindtype, cdbi,
#         colinfo, cref, darkpan, deps, dev, dirtyness, dsn, fixups, fk,
#         fs, hotttnesss, idents, inflator, joinmap, localizing, md, moar,
#         multicolumn, multijoins, normalization, nullability,
#         optimization, overridable, prefetched, rdbms, realiased,
#         recognize, reconnection, recursor, refactored, resultsets,
#         serialized, sqlmaker, subselect, uninserted, varchar, weaklink,
#         weakref
#      3: NoBindVars, ODBC, Postgres, RNO, SQLMaker, braindead, colname,
#         conds, dbh, de, freetds, optimizer, proto, rels, riba, rsrc,
#         scalarrefs, signaled, sqlt, sth, unversioned, wrt
#      4: DBDs, SQLT, attr, ctx, dep, premultiplied, resultsource, savepoint,
#         savepoints, sqla
#      5: DQ, dbic, optimized
#      6: codepath, compat, subq, txn
#      7: SUBOPTIMAL
#      8: LongReadLen, sql, subqueries
#      9: AutoCommit, collapser, cond
#     14: attrs
#     17: SQLA
#     22: resultset
#     26: subquery

[dagolden]

How about "All incorrect words, by number of occurrences:"

[kentfredric]

Now patched and squashed as follows:

# All incorrect words, by number of occurrences:
#      1: ActorRoles, Authorized, AutoCast, BelongsTo, Bloggs, CachedKids,
#         Centos, Compat, DBA, DELETEs, DEV, Easysoft, Eg, FC, FileColumn,
#         FirstSkip, FromForm, HRI, INSERTs, Inflators, LimitXY,
#         LiveObjectIndex, MRO, ManyToMany, Microsft, Multicreate, MyApp,
#         NULLs, NoObjectIndex, Northwind, Optimizer, Overridability,
#         POSTGRESQL, Queryable, RDBMSes, RDMS, REPL, RESULTSET, ReLoad,
#         Reblessing, ResulSet, ResultSetColumn, Resultset, Retrevial,
#         RowNum, SQLT, SchemaVersions, Serializable, Stringfy, Subquery,
#         Subselects, Suported, TMTOWTDI, TxnScopeGuard, UPDATEs, VARCHAR,
#         WhereJoin, YYYY, albumid, artifact, authorization, autodetected,
#         autodetection, autodetects, autoinc, autovalidation, blabla,
#         bugreport, caling, callframe, catalyze, cdbi, chocolateboy,
#         columnname, dbh, dclone, dcloned, dcloning, de, deallocating,
#         deferrable, dsn, explorative, failover, gR, generalized, getter,
#         gravatar, hundres, inflator, initialize, insertdb, instantiation,
#         introspectible, lifecycle, localize, logsystem, materialized,
#         mis, mysqld, nfreeze, nls, noindexed, numification, onwards,
#         optimized, optimizer, organize, params, qw, randomizer,
#         readbound, rebless, recognize, reconnectable, reimplementation,
#         ruleset, scalarref, se, serializable, serialize, serialized,
#         serializing, specialized, sqlt, stacktrace, standardized,
#         suboptimal, superset, sys, sysdate, testdb, thusly,
#         transactionally, txn, uncommited, uninflated,
#         uniqueidentifierstr, unixODBC, unserializable, upto, utilize,
#         versioning, webpages, xyz
#      2: AutoCommit, BUILDARGS, DBs, DESC, EasySoft, FK, FilterColumn,
#         HashRefInflator, Kioku, LimitOffset, LinerNotes, ORMs,
#         PREFETCHING, Postgres, RHS, ResultSets, SELECTs, SkipFirst,
#         Subqueries, attr, autoincremented, autoincrementing, balancers,
#         callsites, colname, ddl, debugcb, fastpath, freeform,
#         incrementing, initialized, nextval, normalize, nullable,
#         overwritable, postgresql, prefetching, recognizes, reconnection,
#         resultsource, serialization, stringifiable, syntaxes, uninserted,
#         varchar
#      3: DBDs, Extensibility, FetchFirst, InflateColumn, Informix, JOINs,
#         PKs, Replicants, RowNumberOver, TT, attrs, behaviors, natively,
#         prefetches, preversion, recognized, subquery, subref, subselects,
#         unary, uniqueidentifier
#      4: DSNs, GenericSubQ, RHEL, Replicant, cond, debugfh, debugobj,
#         overridable, reblessed, subqueries, unversioned
#      5: ORM, ResultSources, callsite, normalizing
#      6: classdata, eg, subselect
#      7: Firebird, prefetched
#      8: savepoints
#      9: Balancer, DDL
#     10: LongReadLen, balancer
#     11: resultsets
#     13: savepoint
#     14: sql
#     15: behavior
#     17: unicode
#     23: ODBC
#     42: replicants
#     43: replicant
#    163: resultset

( NB. Differences in lists of words is because I'm prototyping my changes on a non-test-spelling codebase, so no need for concern, the list with "resultset" as the most common is Test-Spelling's Pod checks, and the one with 'subquery' is extracted from the comments: https://github.com/kentfredric/dbix-class/blob/topic/spelling.t/xt/commentspelling.t#L57 )

[sartak]

Awesome. Thank you @kentfredric! I will ship this in a few days when I return from my trip.

[sartak]

POSTing upload for Test-Spelling-0.20.tar.gz to https://pause.perl.org/pause/authenquery