Default values were not written to NEPMD.INI

[Andreas Schnellbacher]

After installation of NEPMD the EPM doesn't start with NEPMD settings. In contrast, the NEPMD program object 'Standard EPM' starts.

Renaming myepmd\bin\NEPMD.INI is a workaround for that. A new one is created on the next EPM start. The malfunctioning NEPMD.INI contains the entry:

RegDefaults -> \NEPMD\User\MovedUserFiles

while the correct temporary entry would be:

Install -> \NEPMD\User\MovedUserFiles

That explains why the RegDefaults? application wasn't written. It already contains (wrong) data. As a result, EPM won't start.

Additionally, that temporary entry is normally only written if the install script has copied previous files to a backup directory. The backup directory exists with the files etke603.dll and copydll.cmd. Normally they were deleted afterwards.

It seems as if patching etke603.dll to apply the NEPMD colors has failed at the end.

[Andreas Schnellbacher]

I can't reproduce the wrong ini file writing. But the deletion of the temporary files should be fixed. Also another path than myepm\tmp may be suited better.

[Andreas Schnellbacher]

One side effect of the missing default values in RegDefaults? is that EPM doesn't start anymore. It has turned out that only color and font settings are critical.

On initialization, it will now be checked if the query function returns a value. If not, fallback values were applied. That makes the EPM window come up, even in that case.

Partly fixed by r2716. A fix for writing new default values in the case of this bug report will follow.

[Andreas Schnellbacher]

It turns out that ini settings went out of sync between files.

Additionally, the logic to submit info from an installation command to the E macros for the first start had a side effect: Because RegDefaults? still exists, DEFAULTS.DAT/DEFAULTS.CFG was not imported.

The fix includes the setting of default 'emergency' values for ini key queries, even when RegDefaults? is missing. Now the EPM window opens in that case, but only partly initialized. Nevertheless most functions work, also the Save command.

Moreover, the user is now notified about the error and gets the option to work around that.

The fix is mainly contained in r2716 and r2727.

The patching of ETKE603.DLL still has to be reworked and the path logic should be changed, e.g. to use MYEPM\DLL for it instead of replacing the original one. Since there is no real negative impact, I'll close this bug.