Re: fetching files with javascript

What I am taking out of this discussion is:

When using the Lynx keyboard command 'p'(rint) and then its menu choice
'Save to a local file' -- it is inconvenient that the default filename
it supplies includes '?parameters' (and '#frag').

There may be (probably are) other actions within Lynx which similarly
provide a proposed filename derived from a URL.

Request: add to Lynx user options directing it to strip '?params' and
'#frag' from proposed filenames.  Two options so each user can
construct their own preferred behavior.  And, I *think* I would set them
by default to do all the stripping; people could get back the current
behavior by changing the settings.(*)

Another way to handle this would be a new sub-menu: instead of providing
an input buffer pre-filled with the end of the URL, offer selections;
like:

  Saving https://foo.bar.baz/long-obnoxious-name.html?param#frag

  Choose filename:

    1) Type a new name interactively
    2) long-obnoxious-name.html
    3) long-obnoxious-name.html?param#frag
    4) long-obnoxious-name.html?param
    5) long-obnoxious-name.html#frag

(intentionally ordered so #1 is always 'type', #2 is always 'fully
stripped'; #3 only present if param or frag present, then is always
'fully unstripped'; #4 and #5 only appear if input URL has *both* param
and frag.)

There is also a much subtler way this could be implemented.  Currently,
the filename prompt is pre-filled with the last segment of the URL, in
an interactive edit buffer.  And this buffer has history; up-arrow takes
one to the last-saved filename, and so on.  Suppose, then, that the
temporary history buffer for 'foo.html?bar#bar' were pre-filled with the
various stripped forms, *then* followed by actual past history?  For
URLs without ?#, nothing changes at all.  For ones with, reaching real
history is 1-3 extra up-arrows away.  Whatever is selected should then
append to the real history (that is: the proposed alternatives from the
last previous save should *not* leak into the ongoing history buffer).

(*)I'm not sure if there are non-interactive actions which automatically
download URL targets to names derived in this manner.  If yes, it seems
like this would interact badly; e.g. downloading a site recursively and
it has a page offering 'data.cgi?tbl=foo' and 'data.cgi?tbl=bar', which
would then overwrite each other.  In which case, scratch the settings
idea, just do the prompt; only for interactive use.  And possibly add
command line flags directing such strippage, so automatic downloads
could do it *if and when* the caller found it appropriate.

>Bela<