Faré wrote:
I get an all pass on Linux with my hand-compiled clisp.
I have the clisp that you get with Linux Mint, and for me I get this failure:
TEST ABORTED: These two expressions fail comparison with EQUAL: (NEST (LISP-INVOCATION:INVOKE-LISP :CROSS-COMPILE NIL :IMAGE-PATH (NATIVE-NAMESTRING IMG) :EVAL "(uiop:restore-image :entry-point 'hello:entry-point :lisp-interaction nil)" :RUN-PROGRAM-ARGS '(:OUTPUT :LINES :ERROR-OUTPUT T))) evaluates to ("[4]> hello, world") '("hello, world") evaluates to ("hello, world")
This is the same failure I was getting on Mac OS X (and not the same one Dave was getting).
clisp 2.49
As a quick follow-up, when I start clisp the prompt looks like this for me:
[1]>
I am wondering if somehow your hand-built system gets a different prompt or somehow my prompts bleed through into program output in a way yours don't....
Best, r
On Mon, Oct 6, 2014 at 5:12 PM, Robert P. Goldman rpgoldman@sift.info wrote:
Faré wrote:
I get an all pass on Linux with my hand-compiled clisp.
I have the clisp that you get with Linux Mint, and for me I get this failure:
TEST ABORTED: These two expressions fail comparison with EQUAL: (NEST (LISP-INVOCATION:INVOKE-LISP :CROSS-COMPILE NIL :IMAGE-PATH (NATIVE-NAMESTRING IMG) :EVAL "(uiop:restore-image :entry-point 'hello:entry-point :lisp-interaction nil)" :RUN-PROGRAM-ARGS '(:OUTPUT :LINES :ERROR-OUTPUT T))) evaluates to ("[4]> hello, world") '("hello, world") evaluates to ("hello, world")
This is the same failure I was getting on Mac OS X (and not the same one Dave was getting).
clisp 2.49
As a quick follow-up, when I start clisp the prompt looks like this for me:
[1]>
I am wondering if somehow your hand-built system gets a different prompt or somehow my prompts bleed through into program output in a way yours don't....
I don't understand how this error comes to be. I tried with the ubuntu-provided clisp 2.49 (1:2.49-9ubuntu1) instead of the one I custom-built from HEAD a few weeks ago, and it works fine, too. And so I can't debug a problem I can't reproduce and can't fathom.
Can you identify what command is being run? e.g. by tracing run-program in test-program.script. How does the dumped clisp executable image behave without any of the provided arguments? Is there some hidden configuration file that is biting us? Are we failing to disable configuration files when calling the image?
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org If a nation values anything more than freedom, it will lose its freedom; and the irony of it is that if it is comfort or money that it values more, it will lose that, too. — Somerset Maugham (1874–1965)
Faré wrote:
On Mon, Oct 6, 2014 at 5:12 PM, Robert P. Goldman rpgoldman@sift.info wrote:
I don't understand how this error comes to be. I tried with the ubuntu-provided clisp 2.49 (1:2.49-9ubuntu1) instead of the one I custom-built from HEAD a few weeks ago, and it works fine, too. And so I can't debug a problem I can't reproduce and can't fathom.
Can you identify what command is being run? e.g. by tracing run-program in test-program.script. How does the dumped clisp executable image behave without any of the provided arguments? Is there some hidden configuration file that is biting us? Are we failing to disable configuration files when calling the image?
Here's what I see:
$ clisp -M ../build/fasls/clisp-2.49-unix-x64/asdf/test/hello-world-example--all-system\ s.image /usr/lib/clisp-2.49/base/lisp.run: initialization file `../build/fasls/clisp-2.49-unix-x64/asdf/test/hello-world-example--\ all-systems.image' was not created by this version of CLISP runtime rpg@crypto ~/common-lisp/asdf/test $ ../build/fasls/clisp-2.49-unix-x64/asdf/test/hello-world-example--all-systems.image
[1]>
So it seems that these images can't be run as -M arguments to clisp, but they are directly executable.
The [1]> prompt is the same as yours, right? And the [1] is a command count. IIUC there are two mysteries: 1. when I run the test, why does the output include the prompt, instead of being prompt-stripped as it seems to be for you (since the test passes for you)? and 2. why do I have a command count of 4?
R
On Tue, Oct 7, 2014 at 12:14 AM, Robert P. Goldman rpgoldman@sift.info wrote:
Here's what I see:
$ clisp -M ../build/fasls/clisp-2.49-unix-x64/asdf/test/hello-world-example--all-system\ s.image /usr/lib/clisp-2.49/base/lisp.run: initialization file `../build/fasls/clisp-2.49-unix-x64/asdf/test/hello-world-example--\ all-systems.image' was not created by this version of CLISP runtime
That's probably clisp not being able to recognize an image it itself produced if it was made executable. I'd file a bug against clisp if I cared enough.
rpg@crypto ~/common-lisp/asdf/test $ ../build/fasls/clisp-2.49-unix-x64/asdf/test/hello-world-example--all-systems.image
[1]>
So it seems that these images can't be run as -M arguments to clisp, but they are directly executable.
Yes. ASDF, after cl-launch from which it stole dump-image, creates executable images on all implementations that allow it (no override implemented yet — patches welcome I presume), because that can save a lot of headache wrt to always having to distribute two files (executable and image) and/or trust the system to always recompile all images when the implementation is upgraded (something that common-lisp-controller aimed at doing, but ultimately failed because the overhead of requiring debian packaging was too high and lispers wouldn't bother).
The [1]> prompt is the same as yours, right? And the [1] is a command count. IIUC there are two mysteries:
- when I run the test, why does the output include the prompt, instead
of being prompt-stripped as it seems to be for you (since the test passes for you)? and 2. why do I have a command count of 4?
Yes, the [1]> is alright. So my previous hypothesis of an unflushed buffer with [4]> in it seems debunked. I have no other candidate explanation for either of these mysteries. Can you trace run-program and, looking at the command line, tweak it to isolate what causes the issue?
Once again, I cannot reproduce with either the ubuntu 14.04 provided clisp 2.49 or my compilation of the latest 2.49+ from hg (I tried to pull: no change since I last compiled). Could your clisp version be at stake? How recent/old is it? Can you try a recent one?
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org Many who think themselves infinitely superior to the aberrations of Nazism, and sincerely hate all its manifestations, work at the same time for ideals whose realisation would lead straight to the abhorred tyranny. — F.A. Hayek (1899–1992), 'The Road to Serfdom' (1944)
-
Faré -
Robert P. Goldman