CVS log for loncom/lond

Up to [LON-CAPA] / loncom

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN
Current tag: Refactoring

\Q\E-ing the variable in the regular expression

History files/entries were getting made for nohist_ databases becuase like
an idiot I got the sense of the condition backwards.

I liked the logic for sub unsub I put into the main branch better than the
hack job I put in here ... so transplant.

- Fix mis-spelling of unsub in registration.]
- Fix(?) problem with unsub not being able to unsubscribe when resource.ip was not present
  but subscriptions were there in the subscription file.
 (make the subscription file the authoritative information source).
- Add a few little touches to some functions with only one parameter to make interactive
  debugging via telnet easier (telnet terminates lines with \r which chomp won't chomp
  as near as I can tell... I have split even the final parameter from the line with /:/
  so the user can work around this with e.g. unsub:filename: and the filename won't be searched
  for with the \r present on the back end of it.).

Fixed some typos in rolesput

Well the pid was already there to uniquify it.

Append pid to tmp ids to avoid collisions with multiple lond's assigning the same temp id to
different temp files.

- Fix mis-spelling of GetProfileEntryEncrpyted.
- Fix error - if lcuseradd failed, the make user function returned ok
  anyway... very mis-leading.

Mis-spelled ValidateUser as ValidUser in change passwd.

Close security hole in make_passwd_file

Clean compile after refactoring the hell out of the authentication based
functions.

Iron out a few problems.

fix problemwith PasswordPath..extra my masked existing result.

Remove some syntax errors

Fix up some indentation.

Compiles ok with
- Password functions factored out.
- Hash tying functions factored out.
Not tested so don't trust this version!!!

Fix 2 minor errors;
- inc error messages led people to believe the errors were in put.
- in the inc handler, I did not properly reconstruct $userinput for Failure.

Resolve conflicts .

- don't study reg exp, student the string

- retabination extraordinaire

Completed first level break up of lond into 1 handler per request type.
The daemon seems, on the whole to work however:
- There is a lot more testing that must be done.
- There's a lot more potential refactoring to be done.
- There's a need to explore what happens to lonc in the presence
  of lond's disconnecting/exiting.

Corrected some (il)logic having to do with authenticating clients and manager.
This version seems to mostly work but there is a lot of refactoring left to go.
There are also some anomalies between lond/loncnew that have to be examined
when lond exits.   I would not classify this as ready for prime-time.

This version compiles but is known not to work. Refactoring in progress

Correct my mis-manipulation of csv by putting the right version back at the head