[JDEV] main.c

[Thomas Charron]

From: Ted Rolle <ted at acacia.datacomm.com>
Subject: [JDEV] main.c
> The signal processing is promiscuous -- signals other than SIGKILL kill
> it.  Try SIGUSR1, for example.  We may wish to use these signals in the
> future for something else.

  Many signals are reserved and taken care of by pth..  ;-P  The SIGUSR's
are a prime example..

> Some of the file name processing in config.h should be done in main.h.  My
> reasoning for this is that main allows the changing of the configuration
> file via the -c parameter.  So, main.c should "own" the configuration file
> name; parse it, determine its existence, and pass it to config as a
> "validated" file name.  This moves validation code from config.c to main.c

  I'd disagree for one reason.  main.c could 'own' alot of things, but for
the sake of order they are broken out.  In many cases, functionality is
broken out into distinct files according to function, and not 'ownership'.

> -- code which should be executed only once, but with the current design
> executes every time js_config_load is called.  Once jserver starts, the
> file name cannot be changed, therefore it should be parsed at the
> beginning -- and only once.

  Says who?  ;-P  Perhaps mod_admin will allow a reload of an alternate
configuration file, performing a soft restart in the process?  ;-P