On Wed, 8 Mar 2000, Larry W. Virden wrote:
> When running in crawl/traversal mode it would be useful to bypass
> the history list. The reason is that during the process one isn't
> going to use the history list built, and once the history list maximum
> is exceeded one has to bear up with the delay while lynx reports the
> error.
The history list - not its visible representation as "History Page", but
the structure itself that is limited by the maximum - is used by the
traversal logic.
The traversal logic works more or less by simulating keystrokes (RTARROW,
DNARROW, LTARROW) in LYMainLoop.c. There is less difference between
traversal mode and the normal mode than you may think. In particular,
the simulated RTARROW is used to return to the "previous document" -
and the history list is how lynx remembers what the previous document was.
> In fact, it might be useful to turn off error delays during crawl/traversal,
> since user interaction is not available.
There is already
- -nostatus
- -nopause
- setting INFOSECS, MESSAGESECS, and ALERTSECS to 0.
> Finally, I have a structure where I have an index to a fixed series of
> pages and feed this to lynx. For some reason, it appears to be
> looping over and over, rather than stopping once the pages have all
> been encountered. I'm having a difficult time trying to analyze
> what is going wrong here - I thought that the traversal data files
> would ensure that lynx only went into a file once.
If this is after the history maximum has been reached, I wouldn't
be too surprised. The traversal logic seems to assume that the
history list can keep track of the full depth of traversed documents.
Klaus
