Re: lynx-dev Lynx with lss, and reloading options

[Klaus Weide]

On Sun, 19 Dec 1999, Leonid Pauzner wrote:

> 18-Dec-99 11:42 Klaus Weide wrote:
> > I don't remember the exact logic for disabling LYCookieFile changes.
> > Does it still apply, now that there is LYCookieSaveFile?  Does it
> > apply even more?  Does it apply to LYCookieSaveFile?
> 
> The following question should be answered first:
> what to do on reloading if we change LYCookieFile?
> 
>         1) do nothing (till the next session)
>         2) replace LYCookieSaveFile path if it was changed so we
>            save cookies to the new file on exit
>         3) replace LYCookieFile path if it was changed and:
>                 3a) save cookies to the old file
>                 3b) load cookies from the new file
>                 3c) clear cookie jar and load cookies from the new
>                 3d) save cookies to the new file on exit if no
>                       LYCookieSaveFile set
> 
> 
> Perhaps we should do 2), 3b) and 3d).     [currently do 1)]

Changing LYCookieSaveFile automatically (as a result of changing
LYCookieFile) may be counterintuitive and an unexpected surprise.
It makes existence of it as a separate option somewhat pointless.
So no 2), except perhaps if COOKIE_SAVE_FILE is not present at all
(in the new set of options).

Both of 3b) 3c) change the state of the current state management
sessions.  Possibly significantly.  Possibly completely screwing it
up.  These are not things that should happen as a side effect of
changing an option!  They might be appropriate semantics for some
experimental reload-cookie-jar feature.

3a) - saving doesn't belong to LYCookieFile any more but to
LYCookieSaveFile.

3d) - would be the expected consequence of changing LYCookieSaveFile,
not LYCookieFile.


So IMO changing LYCookieFile should continue to have no immediate effect,
except poosible an indirect one of changing LYCookieSaveFile *if that
is not explicitly given*.

LYCookieFile concerns the past, LYCookieSaveFile concerns the future
(as far as the normal actions in the current lynx session go, seen from
the point in time where reload_read_cfg may be called).  Let's not try
to retroactively change the past of a session...


> Another question what to do when we switch persistent_cookies?
>  - from FALSE to TRUE - seems no problem (load the cookie file now)

There are potential problems.  Merging in a set of new cookies, in
mid-session, with the set of session cookies (and cookies
meant-as-persistent that have been treated as session cookies so far)
can result in some confused or undefined state.  Whether that's
particularly likely or not - it should not happen as some side-effect
of changing configuration.  It might be appropriate for some explicit
"load cookie file" or "merge cookie session" etc. function.


>  - from TRUE to FALSE - save cookies now/on exit/never ?

Now: not without prompting, IMO.
On exit: not without prompting, IMO.


   Klaus