CLISP - an ANSI Common Lisp

Hello,

I am trying to use CLISP as the command interpreter for CGI scripts 
under RHEL 6. Basically I created a file called test.cgi with the 
following contents:

=====================
#!/usr/local/bin/clisp -q -norc
(format t "Content-type: text/html~%~%")
(format t "Hello!~%")
=====================

This works perfectly under Ubuntu, while it doesn't under RHEL. The 
webserver error log shows:

=====================
[Thu Jan 03 14:42:52 2013] [error] [client 159.178.28.52] *** -
[Thu Jan 03 14:42:52 2013] [error] [client 159.178.28.52] UNIX error
[Thu Jan 03 14:42:52 2013] [error] [client 159.178.28.52]  13 (EACCES): 
Permission denied
=====================

and after some digging using strace I was able to figure out that the 
problem lies in the following step:

stat("/home/alb/.clisp", 0x7fffbfe6efa0) = -1 EACCES (Permission denied)

The stat() call fails because of selinux policies; if I switch selinux 
to permissive mode then the script works. So for now I have a 
workaround, although not an ideal one, since I would like to switch 
selinux back on if possible.

But my question is: why is CLISP trying to load my own .clisp file? The 
script runs as the apache user, since it's invoked by httpd, the 
test.cgi script is owned by apache, and the clisp executable is in 
/usr/local/bin/ and owned by root. So I can't figure out why CLISP 
thinks that it has to load stuff from my own directory...

Thanks,

Alberto

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Sam Steingold <sds@gn...> - 2013-01-04 15:28 > * Alberto Riva <nyo@...> [2013-01-03 14:38:02 -0500]: > > stat("/home/alb/.clisp", 0x7fffbfe6efa0) = -1 EACCES (Permission denied) > > The stat() call fails because of selinux policies; if I switch selinux > to permissive mode then the script works. So for now I have a > workaround, although not an ideal one, since I would like to switch > selinux back on if possible. the better workaround is to pass "-norc" to clisp. <http://clisp.org/impnotes/clisp.html#opt-norc> > But my question is: why is CLISP trying to load my own .clisp file? > The script runs as the apache user, since it's invoked by httpd, the > test.cgi script is owned by apache, and the clisp executable is in > /usr/local/bin/ and owned by root. So I can't figure out why CLISP > thinks that it has to load stuff from my own directory... clisp takes .clisprc from getenv("HOME"), and, if that is NULL, from getpwnam(getenv("USER")), and, if that does not work, from getpwnam(getlogin()), and, if that does not wor, from getpwuid(getuid()). -- Sam Steingold (http://sds.podval.org/) on Ubuntu 12.04 (precise) X 11.0.11103000 http://www.childpsy.net/ http://camera.org http://thereligionofpeace.com http://palestinefacts.org http://openvotingconsortium.org I am always late to my punctuality class.

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Alberto Riva <alb@ch...> - 2013-01-04 16:32 On 01/04/2013 10:28 AM, Sam Steingold wrote: >> * Alberto Riva <nyo@...> [2013-01-03 14:38:02 -0500]: >> >> stat("/home/alb/.clisp", 0x7fffbfe6efa0) = -1 EACCES (Permission denied) >> >> The stat() call fails because of selinux policies; if I switch selinux >> to permissive mode then the script works. So for now I have a >> workaround, although not an ideal one, since I would like to switch >> selinux back on if possible. > > the better workaround is to pass "-norc" to clisp. > <http://clisp.org/impnotes/clisp.html#opt-norc> Sure, and I did that. This is my CGI script, as I wrote in my original message: ===================== #!/usr/local/bin/clisp -q -norc (format t "Content-type: text/html~%~%") (format t "Hello!~%") ===================== So the first thing I don't understand is actually why CLISP tries to load the (non-existent) ~/.clisp file anyway. >> But my question is: why is CLISP trying to load my own .clisp file? >> The script runs as the apache user, since it's invoked by httpd, the >> test.cgi script is owned by apache, and the clisp executable is in >> /usr/local/bin/ and owned by root. So I can't figure out why CLISP >> thinks that it has to load stuff from my own directory... > > clisp takes .clisprc from getenv("HOME"), > and, if that is NULL, from getpwnam(getenv("USER")), > and, if that does not work, from getpwnam(getlogin()), > and, if that does not wor, from getpwuid(getuid()). So, I checked and the CGI script runs as me instead of the 'apache' user as I would have expected (since it's invoked by httpd). Now I need to figure out why, but this is not a CLISP problem anymore. On the other hand, if the -norc flag worked as expected, I would not have this problem in the first place. One more data point: if I create the .clisp file in my directory and start CLISP, it does *not* get loaded (with or without -norc). So, why does CLISP stat the "~/.clisp" file if then it does not load it? If I instead create .clisprc, then it gets loaded, unless I use -norc, as expected. Alberto

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Zach Beane <xach@xa...> - 2013-01-04 17:01 Alberto Riva <alb@...> writes: > Sure, and I did that. This is my CGI script, as I wrote in my original > message: > > ===================== > #!/usr/local/bin/clisp -q -norc > (format t "Content-type: text/html~%~%") > (format t "Hello!~%") > ===================== > > So the first thing I don't understand is actually why CLISP tries to > load the (non-existent) ~/.clisp file anyway. Shebang lines can only portably handle one argument to the interpreter. Zach

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Alberto Riva <alb@ch...> - 2013-01-04 17:03 On 01/04/2013 11:39 AM, Zach Beane wrote: > Alberto Riva <alb@...> writes: > >> Sure, and I did that. This is my CGI script, as I wrote in my original >> message: >> >> ===================== >> #!/usr/local/bin/clisp -q -norc >> (format t "Content-type: text/html~%~%") >> (format t "Hello!~%") >> ===================== >> >> So the first thing I don't understand is actually why CLISP tries to >> load the (non-existent) ~/.clisp file anyway. > > Shebang lines can only portably handle one argument to the interpreter. I tried removing -q (that is not necessary in this case anyway) leaving only -norc, but I still get the same behavior: [pid 3475] stat("/home/alb/.clisp", 0x7ffff0abe310) = -1 EACCES (Permission denied) Alberto

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Pascal J. Bourguignon <pjb@in...> - 2013-01-04 18:56 Zach Beane <xach@...> writes: > Alberto Riva <alb@...> writes: > >> Sure, and I did that. This is my CGI script, as I wrote in my original >> message: >> >> ===================== >> #!/usr/local/bin/clisp -q -norc >> (format t "Content-type: text/html~%~%") >> (format t "Hello!~%") >> ===================== >> >> So the first thing I don't understand is actually why CLISP tries to >> load the (non-existent) ~/.clisp file anyway. > > Shebang lines can only portably handle one argument to the interpreter. Yes, but IIRC, clisp uses the trick of seeking the beginning of the file when launched to interpret a script, to read the arguments. -- __Pascal Bourguignon__ http://www.informatimago.com/ A bad day in () is better than a good day in {}.

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Sam Steingold <sds@gn...> - 2013-01-04 19:35 > * Alberto Riva <nyo@...> [2013-01-04 11:32:26 -0500]: > >> the better workaround is to pass "-norc" to clisp. >> <http://clisp.org/impnotes/clisp.html#opt-norc> > > Sure, and I did that. This is my CGI script, as I wrote in my original > message: > > ===================== > #!/usr/local/bin/clisp -q -norc > (format t "Content-type: text/html~%~%") > (format t "Hello!~%") > ===================== sorry about that. > So the first thing I don't understand is actually why CLISP tries to > load the (non-existent) ~/.clisp file anyway. me neither. both in my current development tree the ubuntu clisp 2.49: --8<---------------cut here---------------start------------->8--- $ echo '(print 123)' | strace -o z ./clisp -q -norc [1]> 123 123 $ grep clisprc z $ --8<---------------cut here---------------end--------------->8--- > On the other hand, if the -norc flag worked as expected, I would not > have this problem in the first place. indeed. what is the clisp version on RHEL (where it fails)? maybe you could try to debug this? e.g., what does strace (and ltrace) print right before the failure? maybe you could build clisp from the sources and then debug it? <http://clisp.org/impnotes/faq.html#faq-debug> Thanks! -- Sam Steingold (http://sds.podval.org/) on Ubuntu 12.04 (precise) X 11.0.11103000 http://www.childpsy.net/ http://jihadwatch.org http://honestreporting.com http://americancensorship.org http://dhimmi.com http://memri.org There are 10 kinds of people: those who count in binary and those who do not.

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Michael Kappert <michael.kappert@gm...> - 2013-01-04 20:08 On 01/04/2013 05:32 PM, Alberto Riva wrote: > On 01/04/2013 10:28 AM, Sam Steingold wrote: >>> * Alberto Riva <nyo@...> [2013-01-03 14:38:02 -0500]: >>> >>> stat("/home/alb/.clisp", 0x7fffbfe6efa0) = -1 EACCES (Permission denied) >>> >>> The stat() call fails because of selinux policies; if I switch selinux >>> to permissive mode then the script works. So for now I have a >>> workaround, although not an ideal one, since I would like to switch >>> selinux back on if possible. >> >> the better workaround is to pass "-norc" to clisp. >> <http://clisp.org/impnotes/clisp.html#opt-norc> > > Sure, and I did that. This is my CGI script, as I wrote in my original > message: > > ===================== > #!/usr/local/bin/clisp -q -norc > (format t "Content-type: text/html~%~%") > (format t "Hello!~%") > ===================== > > So the first thing I don't understand is actually why CLISP tries to > load the (non-existent) ~/.clisp file anyway. Do you have the common-lisp-controller package installed, and did you install CLisp from a RHEL package? If yes, take a look at /etc/lisp-config.lisp.

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Alberto Riva <alb@ch...> - 2013-01-04 21:23 Attachments: strace.txt      On 01/04/2013 02:34 PM, Sam Steingold wrote: >> * Alberto Riva <nyo@...> [2013-01-04 11:32:26 -0500]: >> >>> the better workaround is to pass "-norc" to clisp. >>> <http://clisp.org/impnotes/clisp.html#opt-norc> >> >> Sure, and I did that. This is my CGI script, as I wrote in my original >> message: >> >> ===================== >> #!/usr/local/bin/clisp -q -norc >> (format t "Content-type: text/html~%~%") >> (format t "Hello!~%") >> ===================== > > sorry about that. > >> So the first thing I don't understand is actually why CLISP tries to >> load the (non-existent) ~/.clisp file anyway. > > me neither. > both in my current development tree the ubuntu clisp 2.49: > > --8<---------------cut here---------------start------------->8--- > $ echo '(print 123)' | strace -o z ./clisp -q -norc > [1]> > 123 > 123 > $ grep clisprc z > $ > --8<---------------cut here---------------end--------------->8--- Please note that the file that causes the problem is called ".clisp", it's NOT .clisprc! If I run the same command you did above and then look in the strace output, I see this: ======================= stat("/home/alb/.clisp", {st_mode=S_IFREG|0777, st_size=24, ...}) = 0 lstat("/home/alb/.clisp", {st_mode=S_IFREG|0777, st_size=24, ...}) = 0 lstat("/home/alb/.clisp", {st_mode=S_IFREG|0777, st_size=24, ...}) = 0 stat("/home/alb/.clisprc", 0x7fff59b52d00) = -1 ENOENT (No such file or directory) lstat("/home/alb/.clisprc", 0x7fff59b53e48) = -1 ENOENT (No such file or directory) stat("/home/alb/.clisprc", 0x7fff59b53c90) = -1 ENOENT (No such file or directory) stat("/home/alb/.clisprc", 0x7fff59b52b50) = -1 ENOENT (No such file or directory) stat("/home/alb/.clisprc.fas", 0x7fff59b52b50) = -1 ENOENT (No such file or directory) stat("/home/alb/.clisprc.lisp", 0x7fff59b52b50) = -1 ENOENT (No such file or directory) stat("/home/alb/.clisprc.lsp", 0x7fff59b52b50) = -1 ENOENT (No such file or directory) stat("/home/alb/.clisprc.cl", 0x7fff59b52b50) = -1 ENOENT (No such file or directory) stat("/home/alb/.clisprc", 0x7fff59b52b90) = -1 ENOENT (No such file or directory) ======================= These calls happen right after reading lispinit.mem. The first stat() is the one that causes my problem. If I rerun with -norc I see this: ======================= stat("/home/alb/.clisp", {st_mode=S_IFREG|0777, st_size=24, ...}) = 0 lstat("/home/alb/.clisp", {st_mode=S_IFREG|0777, st_size=24, ...}) = 0 lstat("/home/alb/.clisp", {st_mode=S_IFREG|0777, st_size=24, ...}) = 0 ======================= In other words, CLISP doesn't look for .clisprc* anymore (correctly), but it still tries to stat the ".clisp" file. >> On the other hand, if the -norc flag worked as expected, I would not >> have this problem in the first place. > > indeed. As the trace above shows, -norc works as expected on the .clisprc file. The .clisp file, instead, seems to be handled differently. > what is the clisp version on RHEL (where it fails)? Red Hat Enterprise Linux Server release 6.3 (Santiago) > maybe you could try to debug this? I would love to, although I'm not very familiar with the CLISP sources. I'll see what I can find out. > e.g., what does strace (and ltrace) print right before the failure? My strace output is attached. The stat() call is on line 223. > maybe you could build clisp from the sources and then debug it? I did build it from source because (also to answer Michael's question) unfortunately it seems there is no RHEL package for CLISP. And I don't have common-lisp-controller. Another piece of information. If I *create* the ~/.clisp file, putting just NIL, or a single PRINT statement in it, and then I try to recompile CLISP from scratch, compilation fails. Specifically, after the first compilation pass, make tries to start lisp.run with the following command: ./lisp.run -B . -N locale -E UTF-8 -Epathname 1:1 -Emisc 1:1 -norc -m 2MW -lp /home/alb/archive/clisp-2.49/src/ -x '(and (load "/home/alb/archive/clisp-2.49/src/init.lisp") (sys::%saveinitmem) (ext::exit)) (ext::exit t)' and I get the following: *** - TRUENAME: pathname with type but without name makes no sense: #P"/home/alb/.clisp" [/home/alb/archive/clisp-2.49/src/eval.d:573] reset() found no driver frame (sp=0x7fff9712bd80-0x7fff971266c0) Exiting on signal 6 Bye. make: *** [interpreted.mem] Aborted (core dumped) Alberto

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Alberto Riva <alb@ch...> - 2013-01-04 22:26 On 01/04/2013 10:28 AM, Sam Steingold wrote: >> >But my question is: why is CLISP trying to load my own .clisp file? >> >The script runs as the apache user, since it's invoked by httpd, the >> >test.cgi script is owned by apache, and the clisp executable is in >> >/usr/local/bin/ and owned by root. So I can't figure out why CLISP >> >thinks that it has to load stuff from my own directory... > clisp takes .clisprc from getenv("HOME"), > and, if that is NULL, from getpwnam(getenv("USER")), > and, if that does not work, from getpwnam(getlogin()), > and, if that does not wor, from getpwuid(getuid()). Regarding this... what's the reason for trying getlogin() *before* getuid()? The getlogin() man page says: "Unfortunately, it is often rather easy to fool getlogin(). Sometimes it does not work at all, because some program messed up the utmp file. Often, it gives only the first 8 characters of the login name. The user currently logged in on the controlling tty of our program need not be the user who started it. Avoid getlogin() for security-related purposes." So it would seem better to try getuid() first. Also, in my specific case CLISP is started by the httpd process that runs as the apache user, but getlogin() returns the username of the user who *started* httpd in the first place, which happens to be me. So, if the code that looks for the ".clisp" file uses the same approach, this would explain why CLISP looks for it in my own directory, which I couldn't understand before. To summarize, my problem is caused by the combination of three factors: a) CLISP tries to stat a file called ".clisp" (even with -norc); b) looking it up in the home directory of the user returned by getlogin() (because the environment that the CGI script runs in does not have HOME and USER); c) and this results in an error *even if the file does not exist* because of SELinux restrictions. We don't know why (a) happens, but I tried switching the order of the getuid() and getlogin() calls in src/pathname.c (in the unix_user_pwd function) and now I get what I think is the correct behavior, ie that the initialization files are looked up in the apache user's home dir rather than mine, and the script works even with SELinux enabled. So, this apparently solves my immediate problem, but I have no idea if it's going to break something else... Alberto

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Sam Steingold <sds@gn...> - 2013-01-06 06:34 > * Alberto Riva <nyo@...> [2013-01-04 16:22:46 -0500]: > > Please note that the file that causes the problem is called ".clisp", > it's NOT .clisprc oops. sorry. this is the value of *user-lib-directory*: http://clisp.org/impnotes/clisp-link.html#opt-install it is a bug that clisp signals an error when it tries to find it. please try this patch: --8<---------------cut here---------------start------------->8--- diff -r 696bfff6e3d1 src/pathname.d --- a/src/pathname.d Mon Dec 03 11:22:27 2012 -0500 +++ b/src/pathname.d Sun Jan 06 01:32:17 2013 -0500 @@ -5753,13 +5753,15 @@ global /*maygc*/ file_kind_t classify_na #endif } -LISPFUNNS(probe_pathname,1) /* (PROBE-PATHNAME pathname) */ -{ /* a safe way to distinguish between files and dirs: +LISPFUN(probe_pathname,seclass_rd_sig,1,0,norest,key,1,(kw(error))) +{ /* (PROBE-PATHNAME pathname &key (error t)) + a safe way to distinguish between files and dirs: "dir", "dir/" ==> #p"dir/" "file", "file/" ==> #p"file" "none", "none/" ==> NIL the first value is the truename, the second is the "correct" absolute pathname */ + bool errorp = !nullp(popSTACK()); if (builtin_stream_p(STACK_0)) { /* stream -> path */ probe_path_from_stream(&STACK_0); /* STACK_0 is now an absolute truename */ } else { /* turn into a pathname */ @@ -5784,6 +5786,10 @@ LISPFUNNS(probe_pathname,1) /* (PROB } }); switch (classification) { + case FILE_KIND_BAD: + if (errorp) + OS_file_error(STACK_0); + /*FALLTHROUGH*/ case FILE_KIND_NONE: /* does not exist */ VALUES1(NIL); skipSTACK(3); return; case FILE_KIND_DIR: { /* directory */ @@ -5828,7 +5834,6 @@ LISPFUNNS(probe_pathname,1) /* (PROB skipSTACK(2); } break; - case FILE_KIND_BAD: OS_file_error(STACK_0); } pushSTACK(asciz_to_string(resolved,O(pathname_encoding))); funcall(L(truename),1); diff -r 696bfff6e3d1 src/spvw.d --- a/src/spvw.d Mon Dec 03 11:22:27 2012 -0500 +++ b/src/spvw.d Sun Jan 06 01:32:17 2013 -0500 @@ -3412,7 +3412,8 @@ local inline void main_actions (struct a /* set *user-lib-directory* */ pushSTACK(ascii_to_string(".clisp/")); pushSTACK(O(user_homedir)); funcall(L(merge_pathnames),2); pushSTACK(value1); /* ~/.clisp/ */ - funcall(L(probe_pathname),1); + pushSTACK(S(Kerror)); pushSTACK(NIL); /* ignore errors */ + funcall(L(probe_pathname),3); if (pathnamep(value1) && nullp(ThePathname(value1)->pathname_name) && nullp(ThePathname(value1)->pathname_type)) diff -r 696bfff6e3d1 src/subr.d --- a/src/subr.d Mon Dec 03 11:22:27 2012 -0500 +++ b/src/subr.d Sun Jan 06 01:32:17 2013 -0500 @@ -828,7 +828,7 @@ LISPFUNN(absolute_pathname,1) LISPFUNNR(namestring,1) LISPFUNNS(truename,1) LISPFUNNS(probe_file,1) -LISPFUNNS(probe_pathname,1) +LISPFUN(probe_pathname,seclass_rd_sig,1,0,norest,key,1,(kw(error))) LISPFUNNS(probe_directory,1) LISPFUNN(delete_file,1) LISPFUN(rename_file,seclass_default,2,0,norest,key,1,(kw(if_exists))) --8<---------------cut here---------------end--------------->8--- -- Sam Steingold (http://sds.podval.org/) on Ubuntu 12.04 (precise) X 11.0.11103000 http://www.childpsy.net/ http://camera.org http://palestinefacts.org http://ffii.org http://mideasttruth.com http://honestreporting.com The biggest childhood mistake is thinking that being an adult is fun.

Re: [clisp-list] Problem with CLISP in cgi script under RHEL From: Sam Steingold <sds@gn...> - 2013-02-05 15:10 did you try the patch? > * Sam Steingold <fqf@...> [2013-01-06 01:33:49 -0500]: > >> * Alberto Riva <nyo@...> [2013-01-04 16:22:46 -0500]: >> >> Please note that the file that causes the problem is called ".clisp", >> it's NOT .clisprc > > oops. sorry. > this is the value of *user-lib-directory*: > http://clisp.org/impnotes/clisp-link.html#opt-install > > it is a bug that clisp signals an error when it tries to find it. > please try this patch: > > diff -r 696bfff6e3d1 src/pathname.d > --- a/src/pathname.d Mon Dec 03 11:22:27 2012 -0500 > +++ b/src/pathname.d Sun Jan 06 01:32:17 2013 -0500 > @@ -5753,13 +5753,15 @@ global /*maygc*/ file_kind_t classify_na > #endif > } > > -LISPFUNNS(probe_pathname,1) /* (PROBE-PATHNAME pathname) */ > -{ /* a safe way to distinguish between files and dirs: > +LISPFUN(probe_pathname,seclass_rd_sig,1,0,norest,key,1,(kw(error))) > +{ /* (PROBE-PATHNAME pathname &key (error t)) > + a safe way to distinguish between files and dirs: > "dir", "dir/" ==> #p"dir/" > "file", "file/" ==> #p"file" > "none", "none/" ==> NIL > the first value is the truename, > the second is the "correct" absolute pathname */ > + bool errorp = !nullp(popSTACK()); > if (builtin_stream_p(STACK_0)) { /* stream -> path */ > probe_path_from_stream(&STACK_0); /* STACK_0 is now an absolute truename */ > } else { /* turn into a pathname */ > @@ -5784,6 +5786,10 @@ LISPFUNNS(probe_pathname,1) /* (PROB > } > }); > switch (classification) { > + case FILE_KIND_BAD: > + if (errorp) > + OS_file_error(STACK_0); > + /*FALLTHROUGH*/ > case FILE_KIND_NONE: /* does not exist */ > VALUES1(NIL); skipSTACK(3); return; > case FILE_KIND_DIR: { /* directory */ > @@ -5828,7 +5834,6 @@ LISPFUNNS(probe_pathname,1) /* (PROB > skipSTACK(2); > } > break; > - case FILE_KIND_BAD: OS_file_error(STACK_0); > } > pushSTACK(asciz_to_string(resolved,O(pathname_encoding))); > funcall(L(truename),1); > diff -r 696bfff6e3d1 src/spvw.d > --- a/src/spvw.d Mon Dec 03 11:22:27 2012 -0500 > +++ b/src/spvw.d Sun Jan 06 01:32:17 2013 -0500 > @@ -3412,7 +3412,8 @@ local inline void main_actions (struct a > /* set *user-lib-directory* */ > pushSTACK(ascii_to_string(".clisp/")); pushSTACK(O(user_homedir)); > funcall(L(merge_pathnames),2); pushSTACK(value1); /* ~/.clisp/ */ > - funcall(L(probe_pathname),1); > + pushSTACK(S(Kerror)); pushSTACK(NIL); /* ignore errors */ > + funcall(L(probe_pathname),3); > if (pathnamep(value1) > && nullp(ThePathname(value1)->pathname_name) > && nullp(ThePathname(value1)->pathname_type)) > diff -r 696bfff6e3d1 src/subr.d > --- a/src/subr.d Mon Dec 03 11:22:27 2012 -0500 > +++ b/src/subr.d Sun Jan 06 01:32:17 2013 -0500 > @@ -828,7 +828,7 @@ LISPFUNN(absolute_pathname,1) > LISPFUNNR(namestring,1) > LISPFUNNS(truename,1) > LISPFUNNS(probe_file,1) > -LISPFUNNS(probe_pathname,1) > +LISPFUN(probe_pathname,seclass_rd_sig,1,0,norest,key,1,(kw(error))) > LISPFUNNS(probe_directory,1) > LISPFUNN(delete_file,1) > LISPFUN(rename_file,seclass_default,2,0,norest,key,1,(kw(if_exists))) -- Sam Steingold (http://sds.podval.org/) on Ubuntu 12.04 (precise) X 11.0.11103000 http://www.childpsy.net/ http://memri.org http://palestinefacts.org http://iris.org.il http://think-israel.org You do not have time or money to sue anyone rich enough to be worth suing.

59 messages have been excluded from this view by a project administrator.