Martin Vermeer wrote:
On Fri, 2006-03-17 at 11:51 +0200, Martin Vermeer wrote:
On Fri, 2006-03-17 at 10:29 +0100, Helge Hafting wrote:
Martin Vermeer wrote:
...
This patch makes
every save intentional, which is a Good Thing.
Lyx already indicates if the document is changed or not.
I may still need to _save_ an unchanged document, because
I know I modified the .lyx file through other means:
* I copied another file over it (most common)
* I edited that .lyx file with another lyx or a text editor (rare)
These are not common situations for most users. "Change" means: this
document in memory has changed. Not: this document in memory is now
different from the stored version on disk. I disagree with you here.
I know that change means "changed in memory".
Now if lyx also detect these other means - go ahead and disable save.
But if mismatch between memory and the lyx file can't be detected
100% reliably, then please don't force us to work around broken
disabling logic.
See above. You expect the change flag to do a job it isn't meant to.
No, I don't need the current change indicator to detect the
weird cases. I usually know when I mess around with a file
that way. But then I might want to save the seemingly
unchanged file - without hassle.
Breaking the (Changed) notifier isn't so bad, as long as I can
save when I know that the lack of (Changed) is wrong.
See above.
There is the easiest of workarounds: do a small edit.
<space><backspace> comes to mind.
Works, but don't want to. Why force people?
It isn't forcing, it's protecting. When writing, clicking 'save' every
now and then becomes second nature. Now when opening a file only for
inspection, you should not be able to accidentally change that file's
date stamp (yes, that's the most hateful thing about re-saving
"unchanged" documents -- and no way within the known laws of physics to
revert it!)
That argument isn't good. I save now and then while writing,
not while inspecting. Someone used to lyx won't save very often
anyway - crashes are few and far between, and even then the
emergency file saves the day. Only developers run a "risky" lyx.
You are trying to protect the user against
her/himself, almost an insult :-) , and adding extra trouble as well.
Sure, the "extra trouble" only ever occur for power users.
That is a corner case. But I wouldn't say "habitual saving accident
while just reading" is a common case either. Or is this what everybody do?
Again, blocking save (with an easy override in the form of a "Touch"
menu entry) is the Right Thing.
A compromise is possible. How about allowing the save button always,
but going the emacs way of responding with "no changes to save"
on the status line?
Then, if I get sufficiently irritated by this, I can write a patch
that checks the timestamp on the file, and does a real save
anyway if the "unchanged" file's timestamp doesn't match
the timestamp on disk. For surely the disk
timestamp changes if the file is updated by means other than lyx.
This approach seems the best of both worlds - you get to
preserve your unchanged timestamps, and I have the option
for a nonintrusive fix for my special case. Well, if I ever
become able to compile lyx again, that is. I still get
that weird namespace error.
Helge Hafting
