#44 After upgrade from 1.0.7.24 to 1.0.8.1: msetup corrupts etc/mareas.data, leaves mareas.temp behind

[lodger]

After upgrading my 1.0.7.24 install to 1.0.8.1 I discovered the following error:

1.) Enter mbsetup option 9.2 (Message Area Setup)
2.) Pick a Netmail area
3.) Set "Days Old" to 3650
4.) Set "Max Msgs." to 9999 -> mbsetup will drop an error message: "unable to write to mareas.data and one is unable to leave mbsetup except for pressing Ctrl+C

5.) In $MBSE_ROOT/etc/ there is then an additional "mareas.temp" file left behind

I can (and have) reproduced this problem multiple times by tarballing my 1.0.7.24 before upgrading and then upgrade to the latest version of the "mbse-code" repo.

Tell me if you need my mareas.data file for further analysis

[lodger]

I did some more testing and this bug seems to affect any type of mail area (Local / Echomail / Netmail), not only Netmail type areas and I can trigger it by trying to change any of the area settings. The moment I do (e.g. change the "Aliases" setting from 'No' to 'Yes' or entering a new "Area Name") mbsetup will display the following error message:

"Unable to open the mareas file, Press any key" - however I can not exit mbsetup at this point except hitting Ctrl + C and forcefully exit the program

in $MBSE_ROOT/etc/ I can then see this:

-rw-rw---- 1 mbse bbs 1488886 Jul 21 10:35 etc/mareas.data
-rw-rw-r-- 1 mbse bbs 1488886 Jul 21 10:35 etc/mareas.temp

[Michael Dillon]

This is related to some duplicate checking code I implemented (though it hadn't had a chance to be fully tested). The error you see is because it failed to open the temp file and stopped, shouldn't be any corruption because the file itself was not loaded. I'm investigating the cause.

[Michael Dillon]

For the time being I have reverted(disabled) the duplication checking code due to some obscure errors. It'll take more time to figure it out but this behavior will be more consistent with previous versions now. It has already been pushed to the repo for now.

[lodger]

Just pulled your latest commit and the issue is gone now. Thanks for providing a workaround that fast!!