newLISP Stable Maintenance Release v.10.4.3

[Lutz]

Version 10.4.3 fixes two critical bugs when working with files.

Files and release notes: http://www.newlisp.org/index.cgi?page=Downloads

[xytroxon]

This code:

Code: Select all

(module "crypto.lsp")
(println (crypto:SHA1 "newLISP"))
(exit)

Crashes on Win 7:

Code: Select all

newLISP v.10.4.3 on Win32 IPv4/6 libffi, execute 'newlisp -h' for more info.

Problem signature:
  Problem Event Name:	APPCRASH
  Application Name:	newlisp.exe
  Application Version:	0.0.0.0
  Application Timestamp:	4fa6cce3
  Fault Module Name:	libeay32.dll
  Fault Module Version:	0.9.8.8
  Fault Module Timestamp:	48bef85f
  Exception Code:	c0000005
  Exception Offset:	000beb50
  OS Version:	6.1.7600.2.0.0.768.3

--xytroxon

[Lutz]

It's (crypto:sha1 "newLISP") with lower-case sha1 not SHA1, the lower level imported function, which is only used internally in the module.

Code: Select all

> (crypto:sha1 "newLISP")
"554f4c8dedd3ff02db493f328793cca0de5987c1"

[xytroxon]

Thanks...

I saw in the "cryptp.lsp" source file:

Code: Select all

(import library "MD5")
(import library "RIPEMD160")
(import library "SHA1")
(import library "SHA256")

... and missed (the next line ;o):

Code: Select all

;; @syntax (crypto:md5 <string> <bool-raw>)

Late night coding and time for larger View panel fonts in FreeCommander!
http://en.wikipedia.org/wiki/FreeCommander

-- xytroxon

P.S.

1. Concern that this minor syntax error caused a system hang / newlisp.exe crash and not just issue a newlisp error message.

2. Would be handy to have all the standard module's html pages pre-installed in the Windows distro.
i.e. accessible from the Start -> All Programs -> newLISP just like "Manual and Reference", "Code Patterns", etc. are. (We Windows users need all the help we can get ;>)

[xytroxon]

Hi Lutz!

This problem about drove me crazy, it only showed up rarely when decoding html base64 data strings...

(My previous late night crypto based "torture test" didn't catch it either ;o)

In PHP, base64 encoding an empty string returns an empty string...

While newLisp returns "===="

Code: Select all

>newlisp
newLISP v.10.4.3 on Win32 IPv4/6 libffi, execute 'newlisp -h' for more info.

> (base64-enc "")
"===="
>

From: RFC 4648
http://tools.ietf.org/html/rfc4648

10. Test Vectors

BASE64("") = "" <---- HELLO!!!

BASE64("f") = "Zg=="

BASE64("fo") = "Zm8="

BASE64("foo") = "Zm9v"

BASE64("foob") = "Zm9vYg=="

BASE64("fooba") = "Zm9vYmE="

BASE64("foobar") = "Zm9vYmFy"

-- xytroxon

[Lutz]

changed in 10.4.4 using an optional flag:

Code: Select all

newLISP v.10.4.4 on OSX IPv4/6 UTF-8 libffi, execute 'newlisp -h' for more info.

> (base64-enc "")
"===="
> (base64-enc "" true)  ; new in version 10.4.4
""
> (base64-dec "")
""
> (base64-dec "====")
""
> 

Ps: the text-based command interface of the guiserver requires the empty string to be encoded as "====" . Although the test-cases in the RFC translate into "", I believe that "====" as encoding for the empty string is still ok. When decoding base64 padding chars "====" back, you get "" again.