newLISP Fan Club

Forum => newLISP in the real world => Topic started by: protozen on June 26, 2014, 09:25:43 PM

Title: newlisp -x and embedding data files
Post by: protozen on June 26, 2014, 09:25:43 PM
Is there a way to embedded files into the compiled executable? I would like to build a self contained website distributed as an executable - delivering static files that have been embedded. Alternatively, can the executable be rewritten if updates were made through such a web app?
Title: Re: newlisp -x and embedding data files
Post by: HPW on June 27, 2014, 01:10:10 AM
Hello,



You may have a look at the link option:



http://www.newlisp.org/downloads/newlisp_manual.html#link



Regards
Title: Re: newlisp -x and embedding data files
Post by: ryuo on June 27, 2014, 11:59:19 AM
If all else fails you can embed the files into the source code as symbols via [text][/text] blocks. You could put these in external modules. newLISP may embed all the modules loaded via the load function, but I am not heavily familiar with how this feature works. As a last resort, you could create a script that builds the full source file from the individual modules before newLISP creates an executable from it. This is easily done with a Makefile, if you've ever used UNIX development tools.
Title: Re: newlisp -x and embedding data files
Post by: protozen on June 27, 2014, 02:34:23 PM
Thanks for the input ryuo, this is what I was thinking I'd have to do if there was no sanctioned way.
Title: Re: newlisp -x and embedding data files
Post by: rrq on June 27, 2014, 06:08:50 PM
I don't think newlisp -x embeds other than the nominated module.

However, you may well use
(write-file "myembedding.lsp" (source))
which gets you a quite decent snapshot of the currently loaded state, in newlisp source.

Then just add the "run clauses", and embed that.
Title: Re: newlisp -x and embedding data files
Post by: rrq on June 27, 2014, 08:17:51 PM
Hmm; I'm sure you'd work it out, but as I tried it, I realized it requires a slightly more complex source clause, such as:
(write-file "myembedding.lsp"
  (join (map source (filter (fn (s) (context? (eval s))) (symbols)))
    ""))

and this needs to be in the MAIN context.
Title: Re: newlisp -x and embedding data files
Post by: rrq on June 28, 2014, 03:24:13 AM
I already had this on my todo list since I also much prefer free standing executables for production code. So, just to complete the thought, I put together a small embedding module, which takes care of including loaded files into the embedding by means of replacing the standard load function. I've made it available here,

//http://www.realthing.com.au/files/newlisp/embed.lsp

It kept me entertained a cold and windy winter Saturday afternoon.
Title: Re: newlisp -x and embedding data files
Post by: protozen on July 08, 2014, 08:31:32 PM
Quote from: "ralph.ronnquist"I already had this on my todo list since I also much prefer free standing executables for production code. So, just to complete the thought, I put together a small embedding module, which takes care of including loaded files into the embedding by means of replacing the standard load function. I've made it available here,

//http://www.realthing.com.au/files/newlisp/embed.lsp

It kept me entertained a cold and windy winter Saturday afternoon.

Hey thanks for this, I appreciate your input and work.
Title: Re: newlisp -x and embedding data files
Post by: Lutz on July 08, 2014, 09:19:50 PM
Also linked from here: http://www.newlisp.org/modules/various/index.html



Ps: changed the location link
Title: Re: newlisp -x and embedding data files
Post by: rrq on July 16, 2014, 04:31:01 AM
I thought to add the note, that often it's best to skip the final embedding step, since that step must be done with the newlisp runtime environment of the target machine. E.g., I have a development machine with libffi.so.6 and a target machine with only libffi.so.5, so my embedded program doesn't run happily ever after.



The point is that the embedded program is still dynamically linked. Therefore it's probably better to rather export the embedding base, which is a self-contained newlisp source, and run this with the target newlisp runtime.  Obviously you should still have the same newlisp version.



Thus, my step 3 should be to add a inital line of
#!/usr/bin/newlisp

(or similar) to the embedding base, to make it an executable script, instead of the -x embedding to make it a dynamic ELF.



Though, maybe one can pre-link a dynamic executable into an exportable static executable (for compatible OS), but I'm not familar with the hoops and incantations for this.