Juan Jose Garcia-Ripoll writes:
> (rename-file "foo.exe" "faa") -> "faa.exe"
>
> Unfortunately at some point it seemed ok to follow SBCL and interpret
> physical pathnames without extension as having :type NIL and drop support
> of :unspecific, as it leads to non-readable pathnames (i.e. a single
Matthew Mondor writes:
> On Tue, 9 Oct 2012 19:55:52 -0400
> Matthew Mondor wrote:
>
>> ECL:
>>
>> (make-package :foo) -> #<"FOO" package>
>> (defparameter foo::gentemp 10) -> GENTEMP
>> (symbol-package 'foo::gentemp) -> #<"COMMON-LISP" package>
>
> Also:
>
> SBCL:
> * (make-package :foo)
>
Juan Jose Garcia-Ripoll writes:
> As you may recall, ECL now ships a module :ecl-quicklisp that installs and
> loads quicklisp. I made several changes to it:
>
> * If quicklisp has already been loaded, ECL will not reload it.
>
> * (require :ecl-quicklisp) installs quicklisp on different location
Juan Jose Garcia-Ripoll writes:
> Anyone has such a function? I would like to plug it into ECL for downloading
> the ANSI test suite.
Quicklisp's http.lisp has something like that, but it's not very short
(it has code to handle chunked encoding for example). The old
asdf-install has http code an
Juan Jose Garcia-Ripoll writes:
> - Standalone executables do not have ASDF in it. There is no need. Yet usocket
> uses it to define a logical hostname which is never used.
ASDF2's enabled-by-default output translations break the practice of
loading file resources relative to *load-truename* or
Matthew Mondor writes:
> On Mon, 18 Jul 2011 13:15:01 -0400
> Matthew Mondor wrote:
>
>> This minor patch adds a third value, the port number. The SBCL
>> documentation on sb-bsd-sockets:socket-accept also only refers to two
>> values, but adding a third value should not hurt existing applicati
Juan Jose Garcia-Ripoll writes:
> On Wed, Mar 2, 2011 at 2:03 AM, Zach Beane wrote:
>
> Say I have a directory structure like this:
>
> /tmp/x/a.txt
> /tmp/x/b.txt (a symlink to a.txt)
>
> Is there any function I can call to get a list of #p"/tmp
"Pascal J. Bourguignon" writes:
> The problem is that CL:DIRECTORY is specified to apply TRUENAME which is
> specified to resolve symbolic links. (See section 20.1.3 Truenames).
>
> There is no conforming way to obtain the pathname of a symbolic link in
> CL. You have to use an extension, eithe
"Duong \"Yang\" Ha Nguyen" writes:
> Hi Zach,
>
> Not sure I understand what you need but have you tried Peter Seibel's
> CL-FAD (CL-FAD:LIST-DIRECTORY)?
Yes, cl-fad does not do what I need.
Zach
--
Free Software Downl
Say I have a directory structure like this:
/tmp/x/a.txt
/tmp/x/b.txt (a symlink to a.txt)
Is there any function I can call to get a list of #p"/tmp/x/a.txt" and
#p"/tmp/x/b.txt? On sbcl, for example, I can pass :resolve-symlinks nil
to CL:DIRECTORY. What (if anything) can I use in ECL?
Than
Juan Jose Garcia-Ripoll writes:
> $ ls -l |grep tmp
> drwxr-xr-x 20 jjgarcia jjgarcia 680 Feb 20 23:59 tmp
> $ ln -sf tmp foo
> $ ecl -norc -eval '(print (truename "foo"))' -eval '(quit)'
>
> #P"/Users/jjgarcia/tmp/"
>
> I would rather use this method. It is more portable and FILE-KIND ma
KDr2 writes:
> Is this one OK?
>
>> (ext:file-kind #P"/" t)
>
> :DIRECTORY
>
> (ext:file-kind path follow-links)
That's perfect, thanks!
Zach
--
Free Software Download: Index, Search & Analyze Logs and other IT data in
If I've gotten a pathname like #p"/foo/bar", is there an ecl-specific
function I can call to determine if "/foo/bar" names a directory?
Thanks,
Zach
--
Free Software Download: Index, Search & Analyze Logs and other IT dat
Márcio Faustino writes:
> Is it easier/faster to use Lisp as a CGI script language under Linux?
A common model for CL web work is to start a long-running Lisp process
and communicate with it over a socket. That way you don't pay the
startup costs repeatedly.
Zach
--
Márcio Faustino writes:
> Hi,
>
> Whenever I install a package via Quicklisp, "xmls" for example, when it is
> compiled the object files (*.fas, *.obj) are placed at:
> C:\Users\marcio\AppData\Roaming\common-lisp\cache\ecl-11.1.1-win-x86\C\Users\marcio\quicklisp\dists\quicklisp\software\s-xml-20
Juan Jose Garcia-Ripoll writes:
> On Mon, Oct 4, 2010 at 2:43 PM, Zach Beane wrote:
>
> I'd like to use Pierre Mai's Deflate library with ECL, but it fails on
> every gzip file I test. Attached is a small tarball that demonstrates
> the problem.
>
>
When Abhishek Reddy told me that older versions of ECL don't fail with
my test case, I tried a git-bisect and found that commit
5cfb33d7ad472e7a70d14069df4e374087c6117c, "Inline forms for
ROW-MAJOR-AREF/ASET acting on specialized arrays with elements
:[u]int{8,16,32,64}-t" fails on the test case, b
Juan Jose Garcia-Ripoll writes:
> Zach, thanks a lot for a self-contained test. Could you please provide us with
> some more information about the platform where you tested ECL? I mean at
> least configuration flags, chipset, 32/64 bits, operating system and value of
> *features*
Sure, sorry for
I'd like to use Pierre Mai's Deflate library with ECL, but it fails on
every gzip file I test. Attached is a small tarball that demonstrates
the problem. It has deflate.lisp from Deflate 1.0.1, a small "gunzip"
definition, and a 107-byte gzipped file input.gz. Loading "fail.lisp"
will compile and l
Juan Jose Garcia-Ripoll writes:
> Hopefully fixed in CVS. RENAME-FILE is all what you need -- even though the
> specification only mandates it should work for files and thus this assumption
> is not portable.
Is there any way to do it that would work in an already-released ECL?
Zach
--
I'd like to rename a directory with ECL. If I have a directory like
"/tmp/from/" and I want to rename it to "/tmp/to/", I get something like
this:
> (rename-file "/tmp/from/" "/tmp/to/")
Cannot rename the file "/tmp/from/" to #P"/tmp/to/".
Explanation: No such file or directory.
Is there a
21 matches
Mail list logo
