lynx-dev Re: [-dev.12] experimental text entry fields patch (updated)

[Kim DeVaughn]

On Tue, Jan 12, 1999, Mike Castle (address@hidden) said:
|
| Amazingly enough Kim DeVaughn said:
| >
| > I wouldn't know about elm or cvs, but trn certainly does (or more
| > accurately, the Pnews and Rnmail scripts).
|
| Strange.  I have left over .article, .article~, .letter, and .letter~
| files.  Then again that is 3.6 on my ISP (I've not built a 4.0 beta in
| quite a while).

And I haven't used trn3.6 for years, and don't recall its particular
idiosyncrasies offhand.  However, from the latest (trn4-test69) Pnews,
we have:

 tmpartcopy=$dotdir/.article
 tmpart=$tmpartcopy.$$
 ...
 trap "$cp $tmpart $tmpartcopy; $rm -f $rmlist $tmpart*; exit \$exitval" 0
                                                      ^
which is the tempfile cleaner-upper.  And unrelated (but indicative of
the "attention to details" mindset), the reason for using "cp" instead
of "mv" in the exit trap, is so as not to break hardlinks to ".article",
if any (yes, there are valid reasons to do that).


| > We did create it.  It's the version of the temp file we created when we
| > wrote the textarea data out to disk, so it could be edited.
|
| I'm not talking about deleting the file that Lynx created.  I'm worried
| about deleting the BACKUP copy an editor may have created.  I don't think
| there should be any attempt to delete THOSE files.  Lynx did not create
| those.  Lynx should not try to delete them.

The logical contents of what ends up being called the backup file (eg,
L12345-3TMP.txt~) *is* what was created by lynx when it dumped out the
textarea data.  Logically, the non-backup file that exists after editing
(eg, L12345-3TMP.txt) was created by you and your editor, not lynx.

I use "logically" in the above to avoid pedantic discussions about which
was the "original" file, and whether a "mv" or a "cp" like operation was
used to create the other copy.  Not to mention that the "duplication"
method may be changed in emacs, a`la:

 (setq backup-by-copying 1)              and/or
 (setq backup-by-copying-when-linked 1)

In the default case, emacs uses a backup-by-renaming type operation, and
the the "backup" not only logically contains the data that lynx itself
wrote, but is exactly the same file, right down to the inode.  Only the
filename was changed during the edit process.

Gee, I guess I did get just a trifle pedantic there, didn't I ... :-) ...?

/kim