Lithper in that lisp.ru forum is "unixtechie" in this one. Here's a translation of Slobin's first comment there:
Главное -- не раскатывать губы. Ньюлисп -- это такой бейсик. Ну или перл времён 4.036 (1993 год, кажется). Очень удобно, если нужен скриптик на сотню строчек в полностью контролируемом вами окружении. И полный афедроlaн, если у вас большой проект и есть возможность кому-то ввести вредоносные данные. SQL injection -- фигня по сравнению с тем, что в ньюлиспе до недавних пор вообще не было аналога функции read -- предлагалось эмулировать её через eval-string (а что при этом выполнится -- что прислали, то и выполнится). О том, что в режиме "сервера приложений" он наивно выполняет вообще всё, что придёт из сокета, я просто молчу. Ну и чисто языковое: ссылки (которые эмулируются через контексты) не являются первоклассными значениями, что делает обработку сколь нибудь сложных структур данных... неудобной.
The main thing is not to hope for too much. Newlisp is a sort of BASIC. Or PErl 4.036 (that was 1993, if I am not mistaken). Very convenient if all you need is a smallish script a hundred lines long in a totally controlled environment. And a total disaster for a large project and if there is a possibility of introducing hostile data. SQL injection is nothing compared to the fact that newlisp until recenty had no "read" function. One was supposed to emulate it with "eval-string" (and what would be executed? whatever was sent over, that's what's executed). Not mentioning that in the "application server" mode New Lisp would naively run everything that is coming from the socket.
And a purely linguistic gripe, too: references (which are emulated with contexts) are not first-class elements of the language, what makes processing of any complex data structures .. unconvenient.
Но для мелких скриптиков -- да, хорош. У меня он практически вытеснил Питон из этой ниши. И, кстати, я таки да раздаю исполняемые файлы наивным пользователям. ;-) И автор, да, поддаётся убеждению, если убеждать настойчиво и аргументированно. Но, боюсencedь, человек, для которого этот язык будет первым, к неряшливому стилю привыкнет на всю жизнь.
But for tiny scriptlets it is OK. It almost pushed python out of my practice in this niche. And, yes, I do hand out packed one-file executables to naive users ;) And yes, newlisp author allows himself to be influenced by others' opinions if one argues logically and persistently.
But someone who starts using NewLisp as his first language //meaning, first Lisp variation -- translator// will get used to an untidy style for life
P.S.
http://slobin.pp.ru/newlisp/
P.P.S. И он не по фамилии Лутц, а по имени Лутц! А по фамилии он Мюллер!
..and Lutz is not his surname. It's his name. His surname is Mu(e)ller
I do not agree, of course, but dropped that discussion thread and did not argure there.
(a) tiny and minimal languages have been used for sizeable projects allright in recent history of IT industry. Remember Lua, or tcl, or even perl (saw a book pushing it as a tool for fast prototyping in large and formal projects in corporate and military environments in the second half of the 90s)
Other examples could be Forth, or another minimal version of Lisp written by a German living in Germany and used by him in his consultancy work etc.
(b) a script bundled with NewLisp as a pseudo-executable that is standalone, can be made protected from unauthorized launching in a network server mode (or any other mode as all input will be filtered by your script directly).
Starting a socket that listens and then executes arbitrary code can be done with "nc" (netcat) or a dozen of other tiny utilities, or from any scripting language installed on your unix box (e.g. perl).
No one seems to be worried too much about it, for some reason. It's a given. The sysadmin should eliminate any such installations from his servers open to the internet, yes.
(c) And I have no patience when someone talks about supposed deficiencies of NewLisp because of its context feature, or reference/copying or macros treatment etc.
First, many other implementations lack something, but no one whines if those guys label themselves as "R4.. or R5.... Scheme" or "Common Lisp" in which "some features are not yet implemented in release xxx"
People are more than happy to bear all crap the windbag Graham feeds them about Ark.
But the whiners are oh so selective when they mention a well-rounded implementation like NewLisp
It's just a double standard.