Hello,
I have recently moved from OpenSSH 3.6 to 3.7 (portable version) :
[pyb at nihao tmp]$ ssh -v
OpenSSH_3.7p1, SSH protocols 1.5/2.0, OpenSSL 0.9.6h 5 Dec 2002
(okay, the SSL release is a little old but that's not the subject).
I am also using scripts to perform automated sftp transfers between some
systems. The 3.7 release of OpenSSH broke my scripts, which made me sad
and my customers angry :-(
The main problem I met is in processing 'ls' output. The format changed
dramatically WRT my scripts' expectations. Where I got ls-like output
previously, now I get :
sftp> ls send
send/. send/..
send/000000676.lotsend/000000677.lotsend/000000678.lot
send/000000679.lotsend/000000680.lotsend/000000681.lotsend/000000682.lotsend/endfile
[my shitty wrapping email client could make the problem not obvious.
Here it is : the file names are concatenated, without spaces in-between]
Known problem ? If so, my apologies. If not, I did a little homework and
found that the sftp-int.c/do_ls_dir function seems to be the culprit. We
have first :
--------
if (!(lflag & SHORT_VIEW)) {
int m = 0, width = 80;
struct winsize ws;
/* Count entries for sort and find longest filename */
for (n = 0; d[n] != NULL; n++)
m = MAX(m, strlen(d[n]->filename));
if (ioctl(fileno(stdin), TIOCGWINSZ, &ws) != -1)
width = ws.ws_col;
columns = width / (m + 2);
columns = MAX(columns, 1);
colspace = width / columns;
}
--------
which aims to compute the column width and column spacing.
But when printing, a few lines below, we perform :
--------
tmp = path_append(path, d[n]->filename);
fname = path_strip(tmp, strip_path);
xfree(tmp);
if (lflag & LONG_VIEW) {
/* irrelevant code */
} else {
printf("%-*s", colspace, fname);
if (c >= columns) {
printf("\n");
c = 1;
} else
c++;
}
--------
What it boils down is that column width and spacing are computed on
filenames, but we print path + filename. So column width is wrong on
almost all cases.
Here is a short patch. Warning : I assumed path and strip_path are never
NULL and that strip_path is never "bigger" thant path. So I did NO
control. This may be a little rough... I bet there are some better ways
out, if this is really an error and not a misunderstanding on my part.
Hth,
--
Pierre-Yves Bonnetain
B&A Consultants - Networks and Computers Security
Phone : +33 (0) 563 277 241 - Fax : +33 (0) 563 277 245
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: openssh-diff
Url:
http://lists.mindrot.org/pipermail/openssh-unix-dev/attachments/20031013/bb412f6a/attachment.ksh