# mw4print ? Known Issues & Research (2026-08-07)

## Symptom

Printing is broken in the current build (post-`b98cb87e`).

- **Without MySQL enabled:** never prints, neither on mission-end (WM_COPYDATA) nor on manual File ⊥ Open.
- **With MySQL enabled:** app locks up ~5 seconds, prints one mission, then never prints again until restart.

"Anymore" = it worked before the MySQL/banner commits.

---

## Print Flow (how it works)

```
Game (MW4.exe)
  戌式 SendMessage(WM_COPYDATA, filename)
       戌式 CMainFrame::OnCopyData
            戌式 CChildView::DoCopyData  ∠ adds CMyPrintInfo to m_lstMPIs queue

1-second WM_TIMER
  戌式 CChildView::OnTimer
       戌式 g_RSF.LoadPR(p->m_strFile)    ∠ opens and parses .pr file
       戌式 g_RSF.DoPrint()               ∠ if nErr==0: remove from queue
       戌式 g_RSF.Reset()
```

Manual path (File ⊥ Open) also adds to `m_lstMPIs`; the timer handles both cases identically.

### DoPrint() internal flow

```
DoPrint()
  LoadPrintParams()               ∠ reads options.ini / banner.txt
  count m_pScoreObjects ⊥ nPlayers
  if nPlayers == 0: return 0 silently (no print, item still removed!)
  PrintDlg(PD_RETURNDEFAULT | PD_RETURNDC)
    戍式 fail ⊥ nErr = -1, CommDlgExtendedError() has the reason
    戌式 ok   ⊥ hDC set
  MakeInfoTable / params.LoadParameters
  [#ifdef MW4PRINT] if g_dbConfig.bEnabled: DB_ExportMatch()   ∠ 5s timeout here
  SetupFonts(DPI)     fail ⊥ nErr = -2
  StartDoc            fail ⊥ nErr = -3
  per-player loop:
    StartPage / DoPrint(hDC, ...) / EndPage   fail ⊥ nErr = -4
  EndDoc (or AbortDoc on error)
  DeleteDC / delete pInfoTable
  return nErr
```

### Key return codes

| nErr | Meaning |
|------|---------|
| 0 | Success (or nPlayers==0 ? silent non-print) |
| -1 | `PrintDlg` returned FALSE ? check `CommDlgExtendedError()` |
| -2 | `SetupFonts` failed |
| -3 | `StartDoc` failed |
| -4 | `EndPage` failed mid-print |

---

## Commits that changed mw4print

| Hash | Description |
|------|-------------|
| `b98cb87e` | **MySQL export, configurable banner text** ? first change |
| `9f3a5044` / `a45be804` | Version bump to 2.0 (rc only) |
| `515adc24` | **persist banner text via banner.txt** ? second change |
| `3256c103` | **-debug diagnostic logging** (current) |

### What `b98cb87e` actually changed inside DoPrint

The ONLY addition inside DoPrint when MySQL is disabled:

```cpp
#ifdef MW4PRINT
    if (g_dbConfig.bEnabled) {
        DB_ExportMatch(...);   // ∠ gated; skipped when bEnabled=false
    }
#endif
```

So with MySQL disabled, `DoPrint()` is byte-for-byte identical to the pre-`b98cb87e` code. If it's failing, the failure is NOT in this block.

### What `515adc24` changed in LoadPrintParams

Fixed a missing backslash in the `options.ini` path:

```cpp
// before: strOI += "options.ini";   ⊥ "C:\MW4options.ini" (wrong)
strOI += "\\options.ini";            ⊥ "C:\MW4\options.ini" (correct)
```

Side-effect: before the fix, all layout params used their hardcoded defaults. After the fix, they read from the real `options.ini`. If that file has unusual values, print layout could differ ? but all params are clamped to sane min/max so they can't cause a crash.

---

## Key facts established

- `MW4PRINT` IS defined ? it's in `mw4dummy.h` (`#define MW4PRINT`), which is `#include`d before everything in `ChildView.cpp`.
- `DB_ExportMatch` is called **after** `PrintDlg`, not before. The 5-second "lockup" with MySQL = connection-timeout inside `DB_ExportMatch`, which is reached only after `PrintDlg` has already succeeded.
- `LoadPR` always returns `true` if the file opens, even if the meta-header parse fails. If parse fails, `m_pScoreObjects = NULL` ⊥ `nPlayers = 0` ⊥ `DoPrint` returns 0 silently.
- `AssetsDirectory1` has no trailing backslash (stripped in `InitInstance`).

---

## Hypotheses explored (and ruled out)

| Hypothesis | Status |
|-----------|--------|
| MySQL blocks before PrintDlg | ? Wrong ? DB_ExportMatch is called *after* PrintDlg |
| MW4PRINT not defined | ? Defined in mw4dummy.h |
| options.ini path fix breaks layout | ? All params clamped; can't cause crash |
| Message-map corruption from new dialogs | ? Maps look correct |
| LoadPR always fails | ? Returns true if file exists, regardless of parse |
| Memory corruption from DB code | Unconfirmed |
| PrintDlg returns FALSE (no default printer) | **Active hypothesis** |
| nPlayers == 0 (parse failure / all bots) | **Active hypothesis** |

---

## Current debugging approach

Commit `3256c103` adds a `-debug` flag. Run:

```
mw4print.exe -debug
```

Output goes to `mw4print-debug.txt` next to the exe (append, timestamped).

### What the log tells us

```
DoPrint: nPlayers=0
```
⊥ File opened but no players found. Either the .pr file has an unrecognised meta header,
  or all players are bots and `g_bPRTest=false` (no "Include Bot" checkbox ticked).

```
DoPrint: PrintDlg FAILED nErr=-1, CommDlgExtendedError=0000100A
```
⊥ `CDERR_NODEFAULTPRN` ? no default printer is configured on this machine.

```
DoPrint: PrintDlg FAILED nErr=-1, CommDlgExtendedError=00000000
```
⊥ PrintDlg returned FALSE with no error. Unusual; may indicate a hidden dialog or
  modal conflict.

```
OnTimer: LoadPR FAILED for 'C:\...'
```
⊥ `fopen` failed ? file was deleted between WM_COPYDATA arriving and the timer firing.

---

## Next steps

1. Build from current source on Windows (quick ? only `ChildView.cpp` recompiles).
2. Run `mw4print.exe -debug`, do File ⊥ Open on a known .pr file, wait 2 seconds.
3. Read `mw4print-debug.txt` and look for the first failure line.
4. Act on the specific error:
   - `nPlayers=0` ⊥ check .pr file validity; check if "Include Bot" is needed
   - `CommDlgExtendedError=0000100A` ⊥ configure a default printer on the test machine
   - Other ⊥ investigate that specific failure path