variables in load paths

[Brandon Barker]

Sorry if it has been asked before, I will try to update the approrpiate docs as needed. If you have something like:

staload JNI = "{$JNI}/SATS/jni.sats"

How do you set $JNI? Environment variables seems to not be the answer.

[Hongwei Xi]

You can set it as follows:

#define JNI_targetloc "$PATSHOMERELOC/contrib/JNI"

--
You received this message because you are subscribed to the Google Groups "ats-lang-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-lang-user...@googlegroups.com.
To post to this group, send email to ats-lan...@googlegroups.com.
Visit this group at http://groups.google.com/group/ats-lang-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/10bef205-098c-48c0-85ba-f07b0ab6f76b%40googlegroups.com.

[Brandon Barker]

I defined this in e.g Gradle/Calculator.dats (trying several permutations of fulling expanding $PATSHOMERELOC and setting within %{^ %}) to no avail.

[gmhwxi]

You can also do it this way:

patsopt -DATS JNI_targetloc=\"${JNI}\" ...

if you set JNI as an environment variable.

[Brandon Barker]

Oh, maybe I need to use atspkgreloc... 

[gmhwxi]

Maybe PATSHOMERELOC was not properly set.

On Wednesday, December 10, 2014 12:55:10 PM UTC-5, Brandon Barker wrote:

[Brandon Barker]

It seems to be correct in the environment:

echo $PATSHOMERELOC 

/home/brandon/ATS-Postiats-contrib

To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/70ed9a31-c0db-43d4-b217-e6dbff5b2824%40googlegroups.com.

--

Brandon Barker
brandon...@gmail.com

[gmhwxi]

If you paste your code here, then we should be able to find out.

[Brandon Barker]

I tried this one too, but it seems to be the same (I have attached the file):

brandon@brandon-OEM ~/ATS-Postiats-contrib/projects/SMALL/Calculator/Gradle $ echo $JNI
/home/brandon/ATS-Postiats-contrib/contrib/JNI
brandon@brandon-OEM ~/ATS-Postiats-contrib/projects/SMALL/Calculator/Gradle $ patsopt -DATS JNI_targetloc=\"${JNI}\" -d Calculator.dats  | grep JNI
/home/brandon/ATS-Postiats-contrib/contrib/JNI/SATS/jni.sats: 249(line=15, offs=1) -- 284(line=17, offs=3)
#include "JNI/CATS/jni.cats"
ATSdyncst_mac(atscntrb_JNI_GetStringUTFChars)
ATSdyncst_mac(atscntrb_JNI_ReleaseStringUTFChars)
ATSdyncst_mac(atscntrb_JNI_RaiseExceptionByClassName)
ATSINSflab(__patsflab_JNI_eval):
ATSINSmove(tmp960, atscntrb_JNI_GetStringUTFChars(arg0, arg2)) ;
ATSINSmove_void(tmp962, atscntrb_JNI_ReleaseStringUTFChars(arg0, arg2, tmp960)) ;
ATSINSmove_void(tmp964, atscntrb_JNI_RaiseExceptionByClassName(arg0, ATSPMVstring("java/lang/IllegalArgumentException"), ATSPMVstring("ParsingError"))) ;

[Hongwei Xi]

If you can already generate *_dats.c, it means the following line worked:

staload JNI = "{$JNI}/SATS/jni.sats"

--
You received this message because you are subscribed to the Google Groups "ats-lang-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-lang-user...@googlegroups.com.
To post to this group, send email to ats-lan...@googlegroups.com.
Visit this group at http://groups.google.com/group/ats-lang-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/a3e802b0-ee6d-4d00-9746-433624984b6d%40googlegroups.com.

[gmhwxi]

I tried your source code and encountered no problems.

By the way, JNI_targetloc is already defined in share/atspre_define.hats:

https://github.com/githwxi/ATS-Postiats/blob/master/share/atspre_define.hats

There is no need to define it again unless your new definition is different.

On Wednesday, December 10, 2014 1:29:51 PM UTC-5, gmhwxi wrote:

If you can already generate *_dats.c, it means the following line worked:

staload JNI = "{$JNI}/SATS/jni.sats"

[Brandon Barker]

I think I may have spotted a problem but need to step out for a few minutes:

  "${INCLUDE} -IIATS ${PATSHOME}/contrib -I${JNIincl} -c Calculator.dats"

} // end of [Calculator_dats_o]

I think it should be reloc; do you have a symlink there to ATS-contrib/contrib?

--
You received this message because you are subscribed to the Google Groups "ats-lang-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-lang-user...@googlegroups.com.
To post to this group, send email to ats-lan...@googlegroups.com.
Visit this group at http://groups.google.com/group/ats-lang-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/e7fbc06e-c597-4f42-b473-acc81e664371%40googlegroups.com.

--

Brandon Barker
brandon...@gmail.com

[gmhwxi]

You got it :)

Please change PATSHOME to PATSHOMERELOC.

In my setting, these two are the same.

[gmhwxi]

FYI. I have fixed it in the Github.

[Brandon Barker]

I seem to be having a bit of trouble with this mechanism still. The #define works (commented out now), but the command line option shown below must not be quite right:

brand_000@bbsurfacepro3 ~

$ patscc -DATS UNSAFE_targetloc="/home/brand_000/ATS-Postiats/prelude/SATS" hello.dats

patsopt: error(0): the command-line argument [UNSAFE_targetloc=/home/brand_000/ATS-Postiats/prelude/SATS] cannot be properly parsed.

exit(ATS): uncaught exception: _2home_2brand_000_2ATS_2dPostiats_2src_2pats_error_2esats__FatalErrorExn(1025)

brand_000@bbsurfacepro3 ~

$ patscc -DATS UNSAFE_targetloc=\"/home/brand_000/ATS-Postiats/prelude/SATS\" hello.dats

patsopt: error(0): the command-line argument [UNSAFE_targetloc=/home/brand_000/ATS-Postiats/prelude/SATS] cannot be properly parsed.

exit(ATS): uncaught exception: _2home_2brand_000_2ATS_2dPostiats_2src_2pats_error_2esats__FatalErrorExn(1025)

//#define UNSAFE_targetloc "$PATSHOME/prelude/SATS"

staload UN = "{$UNSAFE}/unsafe.sats"

implement

main0 () = {

  val x = $UN.cast2size(7)

  val () = println! ("{$UNSAFE}")

  val () = println! ("Hello world!") // English

  val () = println! x

} (* end of [main0] *)

On Wed, Dec 10, 2014 at 12:56 PM, gmhwxi <gmh...@gmail.com> wrote:

You can also do it this way

patsopt -DATS JNI_targetloc=\"${JNI}\" ...

if you set JNI is as an environment variable.

On Wednesday, December 10, 2014 12:33:37 PM UTC-5, gmhwxi wrote:

To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/7df46954-3f6f-48e4-999d-bf0c448f7f34%40googlegroups.com.

--

Brandon Barker
brandon...@gmail.com

[Hongwei Xi]

Try:

patscc -DATS UNSAFE_targetloc=\\\"/home/brand_000/ATS-Postiats/prelude/SATS\\\" hello.dats

To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/CAORbNRpYFTW3_JQapuWQr9zyQ6j%3D6rXH52iKeQ%3DhxJ2avB%3D60w%40mail.gmail.com.

[Brandon Barker]

Ah yes, this strange cygwin quoting issue again. I'll make a note on the wiki which I'll hopefully find in case I forget again.

Thanks!

To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/CAPPSPLpa1rsSj1HTt6cKvevC19LhTMepJ2TUxVKmXbrc-tceUQ%40mail.gmail.com.

--

Brandon Barker
brandon...@gmail.com

[Yannick Duchêne]

Le mercredi 10 décembre 2014 18:33:37 UTC+1, gmhwxi a écrit :

You can set it as follows:

#define JNI_targetloc "$PATSHOMERELOC/contrib/JNI"

I have three related questions.

Are underscore formally allowed in path variable names? No path variable use any underscore in the distribution nor in contrib, so I wonder.

What explicitly if a variable is unknown? it is to be left as is or it is to be substituted an empty string?

Are substitutions to be recursive?

I know I could play with patscc to test, this is just that i prefer authoritative answers.

[gmhwxi]

1. The character '_' can be used in path names
2. Say you use $FOO but FOO is not defined (or defined as a non-string), the $FOO changes to FOO.
3. Recursive substitution is not supported (it can be easily supported if needed).

[Yannick Duchêne]

Le lundi 28 décembre 2015 04:36:57 UTC+1, gmhwxi a écrit :

1. The character '_' can be used in path names

Path variable names, I assume.

 

2. Say you use $FOO but FOO is not defined (or defined as a non-string), the $FOO changes to FOO.

Okay.

 

3. Recursive substitution is not supported (it can be easily supported if needed).

 That's OK, not an issue. This is just for a tool which needs to be able to locate file the same way Postiats do, and it was worth to ask for documentation too.

[gmhwxi]

I changed the implementation a bit to support recursive substitution;
the maximal recursion depth is set to be 100.