appendhistory and history file truncation

appendhistory and history file truncation

[Danek Duvall]

I haven't seen this show up on the list, so I thought I'd chime in.

I'm having a problem wherein whenever I start up an invocation of zsh
(2.6 beta10), my history file gets wiped.  It doesn't happen when
NO_RCS is set (if I do zsh -f).  It seems that even though
APPENDHISTORY is set, savehistfile() ends up opening the file mode
O_TRUNC on line 1471 of hist.c (at least, I think this is the
culprit).  Perhaps this line should be rewritten as the block in lines
1420-1423.

I checked beta11-test10, and it hasn't been fixed.

Here's a really straightforward patch to beta10, which *should* fix
the problem, though I haven't tested it.  Rewrite as you wish, but I
think this is the right thing to do.

Thanks,
Danek

--- hist.c.orig	Sat Oct 21 03:54:08 1995
+++ hist.c	Sat Oct 21 03:56:30 1995
@@ -1468,7 +1468,11 @@
 		histnum++;
 	    }
 	    fclose(out);
-	    if ((out = fdopen(open(s, O_WRONLY | O_TRUNC, 0600), "w"))) {
+	    if (app & 1)
+		out = fdopen(open(s, O_CREAT | O_WRONLY | O_APPEND, 0600), "a");
+	    else
+		out = fdopen(open(s, O_CREAT | O_WRONLY | O_TRUNC, 0600), "w");
+	    if (out) {
 		if (histnum < savehist)
 		    for (i = 0; i < histnum; i++)
 			fprintf(out, "%s", store[i]);

-- 
Danek Duvall <duvall@lorien.princeton.edu>    For my PGP key, send mail with
Computer Science, Princeton University        subject "get pgp key" or look
http://lorien.princeton.edu/~duvall/          it up on the keyservers.

Re: appendhistory and history file truncation

[Mark Borges]

>> On Sat, 21 Oct 1995 03:58:32 -0400,
>> Danek Duvall(DD) wrote:
DD> I'm having a problem wherein whenever I start up an invocation of zsh
DD> (2.6 beta10), my history file gets wiped.  It doesn't happen when
DD> NO_RCS is set (if I do zsh -f).  It seems that even though
DD> APPENDHISTORY is set, savehistfile() ends up opening the file mode
DD> O_TRUNC on line 1471 of hist.c (at least, I think this is the
DD> culprit).  Perhaps this line should be rewritten as the block in lines
DD> 1420-1423.

DD> I checked beta11-test10, and it hasn't been fixed.

Zoltan and Peter have sent fixes for this in

	archive/latest/373 (also in 337)
	archive/latest/376 

respectively.

Please try to incorparate these in beta11, Richard. For those of us
who are exec'ing the latest betas out of a startup script it's very
unpleasant to lose your history each time you log in. It's very
difficult to do my real work without history, consequently I haven't
been stress testing the baseline betas since this bug appeared.

  -mb-

Re: appendhistory and history file truncation

[Zvi Har'El]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=none, Size: 598 bytes --]

On Sat Oct 21 17:27:24 1995, Mark Borges wrote about ``Re: appendhistory and history file truncation'':
> 
> Zoltan and Peter have sent fixes for this in
> 
> 	archive/latest/373 (also in 337)
> 	archive/latest/376 
> 
Forgive my ignorance, but which ftp site these refer to?

-- 
Dr. Zvi Har'El <rl@math.technion.ac.il>              Department of Mathematics
+972-4-294094(Phone)                 Technion - Israel Institute of Technology
+972-4-324654(FAX)     http://gauss.technion.ac.il/~rl     Haifa 32000, ISRAEL
``If you can't say somethin' nice, don't say nothin' at all.''--Thumper (1942)

Re: appendhistory and history file truncation

[Zvi Har'El]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=none, Size: 2116 bytes --]

On Sat Oct 21 09:58:32 1995, Danek Duvall wrote about ``appendhistory and history file truncation'':
> 
> I haven't seen this show up on the list, so I thought I'd chime in.
> 
> I'm having a problem wherein whenever I start up an invocation of zsh
> (2.6 beta10), my history file gets wiped.  It doesn't happen when
> NO_RCS is set (if I do zsh -f).  It seems that even though
> APPENDHISTORY is set, savehistfile() ends up opening the file mode
> O_TRUNC on line 1471 of hist.c (at least, I think this is the
> culprit).  Perhaps this line should be rewritten as the block in lines
> 1420-1423.
> 
> I checked beta11-test10, and it hasn't been fixed.
> 
> Here's a really straightforward patch to beta10, which *should* fix
> the problem, though I haven't tested it.  Rewrite as you wish, but I
> think this is the right thing to do.
> 
> Thanks,
> Danek
> 
> --- hist.c.orig	Sat Oct 21 03:54:08 1995
> +++ hist.c	Sat Oct 21 03:56:30 1995
> @@ -1468,7 +1468,11 @@
>  		histnum++;
>  	    }
>  	    fclose(out);
> -	    if ((out = fdopen(open(s, O_WRONLY | O_TRUNC, 0600), "w"))) {
> +	    if (app & 1)
> +		out = fdopen(open(s, O_CREAT | O_WRONLY | O_APPEND, 0600), "a");
> +	    else
> +		out = fdopen(open(s, O_CREAT | O_WRONLY | O_TRUNC, 0600), "w");
> +	    if (out) {
>  		if (histnum < savehist)
>  		    for (i = 0; i < histnum; i++)
>  			fprintf(out, "%s", store[i]);
> 
> -- 
> Danek Duvall <duvall@lorien.princeton.edu>    For my PGP key, send mail with
> Computer Science, Princeton University        subject "get pgp key" or look
> http://lorien.princeton.edu/~duvall/          it up on the keyservers.
> 
> 
> 
As far as I have been able to verify, this works for non-login shells. However,
if I log out and then login again, the problem persists even after applying the
patch.

-- 
Dr. Zvi Har'El <rl@math.technion.ac.il>              Department of Mathematics
+972-4-294094(Phone)                 Technion - Israel Institute of Technology
+972-4-324654(FAX)     http://gauss.technion.ac.il/~rl     Haifa 32000, ISRAEL
``If you can't say somethin' nice, don't say nothin' at all.''--Thumper (1942)

Re: appendhistory and history file truncation

[Zoltan Hidvegi]

Zvi Har'El wrote:
> > 	archive/latest/373 (also in 337)
> > 	archive/latest/376 
> > 
> Forgive my ignorance, but which ftp site these refer to?

These are the article numbers in the zsh-workers maining list.  To retrieve an
article send a message to zsh-workers-request@math.gatech.edu with archive as
the subject.  The message should contain something like

get latest/373

An other note: please get your site administrator configure the elm program
you use.  Your message came with this header:

Content-Type: text/plain; charset=none
Content-Transfer-Encoding: 8bit

My elm thinks that this message needs metamail processing.  I tried to add
none to compatcharsets but this did not help.

Cheers,
  Zoltan