Files

Ken Johnson d43f996604 Wildcat: full SDK init (InitWCglobal + LoadMakeWild + BTInitIsam); test across 7 conferences

The TWildcatBase.OpenConference was creating a TMsgDatabase without
populating the SDK globals, causing access violations on MwConfig.
Now calls the full Register sequence: InitWCglobal, LoadMakeWild,
BTInitIsam before opening; BTExitIsam + DisposeWCglobal on close.

test_wildcat reads conferences 0..6 from vendored testdata (copied
to /tmp/ma_wildcat so the source tree stays read-only).

2026-04-15 08:41:02 -07:00

3.4 KiB

Raw Blame History

External dependencies in copied format units

Sample data

Real message bases for testing live at ~/fidonet/msg/ and for WildCat! at ~/allfix_dev/f_drive/tpfiles/allfix/wc_dev/testdata/:

hudson/ — Hudson MSGINFO/IDX/HDR/TXT/TOIDX.BBS set
jam/ — many JAM echo areas (e.g. 10thamd.jhr/jdt/jdx/jlr)
echomail.jam/ — extra JAM samples
netmail/ — FTS-1 *.MSG numbered files (mixed upper/lower case on disk)
squish/ — Squish bases (currently empty but layout reserved)
local/jam/, lowerit/, passthru/ — additional collections
WildCat 4 testdata:
- CONFDESC.DAT/IX/UX — conference descriptors (7 conferences)
- MSG/MSG0..MSG6.DAT + .IX — per-conference message databases
- DATA/ALLUSERS.DAT + USERCONF.DAT — users + per-user conference membership

Tests should open these read-only and verify message counts, first/last messages, and specific known attributes. Do NOT write to these paths from tests — copy them into a scratch dir first. The Wildcat SDK mutates the message database during Open (btis locks, modcounters) so even "read" paths must go through a scratch copy.

Authoritative specs

The FTSC document collection at /home/ken/Source Code/ftsc/docs/ is the source of truth when validating attribute bits, packet header layouts, date strings, and kludge syntax. Key documents we cross-check against:

fts-0001.016 — FTS-1 packet & message format (PKT, *.MSG, attributes, date string, kludges)
fsc-0039.001 / fsc-0039.004 — Type-2+ packet header
fsc-0045.001 — Type-2.2 packet header
fsc-0048.002 — AuxNet Type-2 variant
fsc-0009, fsc-0035 etc. — kludges and control lines

When in doubt about an encoding or bit position, reference the spec before changing the conversion tables in ma.types.pas.

These references were inherited from the Allfix tree and need to be resolved during the format-by-format refactor pass.

ma.fmt.pkt.pas

Currently uses: msgunit, netaddrs, aftypes.

Needs from msgunit:

PktHeader — 58-byte FTN packet header record
PktRecord — per-message header record inside a packet
PacketRecord — high-level parsed message with kludges, address parts, flavours, etc.
FlavourType, FlavourTypeSet, DateTimeArray

Plan:

Move PktHeader + PktRecord into ma.fmt.pkt.pas itself (packet-only).
Move PacketRecord, FlavourType, FlavourTypeSet, DateTimeArray into ma.types.pas so other backends can use them where needed.
Drop the msgunit / netaddrs / aftypes uses clauses.

ma.fmt.wildcat.pas + ma.fmt.wcutil.pas

Use the WildCat 4 SDK. The SDK source has been copied verbatim into src/wc_sdk/ (formerly wc_dev/ in Allfix) so this repo can build the Wildcat backend without external paths.

The SDK pulls in OS/2-era helpers (Os2Comp, DosUtil, p_System, btisbase, isamtool, …). Keeping it on its own search-path entry (-Fusrc/wc_sdk) means callers that don't need Wildcat can omit the backend without forcing the SDK onto the build line.

include.inc

Allfix sources start with {$I include.inc}. Our copied units no longer have that on the search path (we deliberately want the new lib to be self-contained). Plan during refactor:

Replace each {$I include.inc} with explicit {$mode objfpc}{$H+} plus any other compiler switches the unit actually needs (most don't need more than mode + $H+).
Drop WILDCAT4, EXECSWAP, UNI and other Allfix-specific defines.

3.4 KiB Raw Blame History