Re: [fricas-devel] default arguments to sbcl, again
On Sun, Apr 14, 2024 at 07:50:02PM +0200, Waldek Hebisch wrote: > On Sun, Apr 14, 2024 at 06:31:11PM +0200, Waldek Hebisch wrote: > > On Sun, Apr 14, 2024 at 11:03:36AM +0800, Qian Yun wrote: > > > diff --git a/configure.ac b/configure.ac > > > index 073f3e96..0c66c8de 100644 > > > --- a/configure.ac > > > +++ b/configure.ac > > > @@ -452,6 +452,8 @@ case $fricas_lisp_flavor in > > > sbcl) > > > fricas_quiet_flags='--noinform --noprint --no-sysinit > > > --no-userinit' > > > fricas_eval_flags='--eval' > > > + sbcl_memory_size=`echo "(princ (sb-ext:dynamic-space-size))" | > > > $fricas_lisp $fricas_quiet_flags` > > > + test $sbcl_memory_size -gt 2147483647 || > > > fricas_quiet_flags="--dynamic-space-size 2048 $fricas_quiet_flags" > > > ;; > > > cmucl) > > > fricas_quiet_flags='-batch -nositeinit -noinit' > > > > > > > > > This should work. > > > > AFAICS it does not work. Fails with defualt 2.4.3. RSIMP is > > compiled by 'interpsys' and 'sb-ext:dynamic-space-size' is not > > propagated to inteprsys. > > Oops, it seems that dynamic space size size is propagated, but > that 2Gb is not enough for sbcl-2.4.3 with current trunk. Using 3G (that is increasing number in your patch) works. > > -- > Waldek Hebisch > > -- > You received this message because you are subscribed to the Google Groups > "FriCAS - computer algebra system" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to fricas-devel+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/fricas-devel/ZhwXSsLnufSUbO7U%40fricas.org. -- Waldek Hebisch -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/ZhwiN7GH5PDOSkU0%40fricas.org.
Re: [fricas-devel] default arguments to sbcl, again
On Sun, Apr 14, 2024 at 06:31:11PM +0200, Waldek Hebisch wrote: > On Sun, Apr 14, 2024 at 11:03:36AM +0800, Qian Yun wrote: > > diff --git a/configure.ac b/configure.ac > > index 073f3e96..0c66c8de 100644 > > --- a/configure.ac > > +++ b/configure.ac > > @@ -452,6 +452,8 @@ case $fricas_lisp_flavor in > > sbcl) > > fricas_quiet_flags='--noinform --noprint --no-sysinit --no-userinit' > > fricas_eval_flags='--eval' > > + sbcl_memory_size=`echo "(princ (sb-ext:dynamic-space-size))" | > > $fricas_lisp $fricas_quiet_flags` > > + test $sbcl_memory_size -gt 2147483647 || > > fricas_quiet_flags="--dynamic-space-size 2048 $fricas_quiet_flags" > > ;; > > cmucl) > > fricas_quiet_flags='-batch -nositeinit -noinit' > > > > > > This should work. > > AFAICS it does not work. Fails with defualt 2.4.3. RSIMP is > compiled by 'interpsys' and 'sb-ext:dynamic-space-size' is not > propagated to inteprsys. Oops, it seems that dynamic space size size is propagated, but that 2Gb is not enough for sbcl-2.4.3 with current trunk. -- Waldek Hebisch -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/ZhwXSsLnufSUbO7U%40fricas.org.
Re: [fricas-devel] default arguments to sbcl, again
On Sun, Apr 14, 2024 at 11:03:36AM +0800, Qian Yun wrote: > diff --git a/configure.ac b/configure.ac > index 073f3e96..0c66c8de 100644 > --- a/configure.ac > +++ b/configure.ac > @@ -452,6 +452,8 @@ case $fricas_lisp_flavor in > sbcl) > fricas_quiet_flags='--noinform --noprint --no-sysinit --no-userinit' > fricas_eval_flags='--eval' > + sbcl_memory_size=`echo "(princ (sb-ext:dynamic-space-size))" | > $fricas_lisp $fricas_quiet_flags` > + test $sbcl_memory_size -gt 2147483647 || > fricas_quiet_flags="--dynamic-space-size 2048 $fricas_quiet_flags" > ;; > cmucl) > fricas_quiet_flags='-batch -nositeinit -noinit' > > > This should work. AFAICS it does not work. Fails with defualt 2.4.3. RSIMP is compiled by 'interpsys' and 'sb-ext:dynamic-space-size' is not propagated to inteprsys. > BTW, I should update configure with which autoconf, 2.71? Next update to configure should be with autoconf 2.71. > Regarding "--disable-ldb --disable-debugger", they should not be > used when dumping lisp image, instead they can be used when batch > compiling lisp files. OK, that is reasonable. -- Waldek Hebisch -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/ZhwEzxhkD8NLjBWj%40fricas.org.
Re: [fricas-devel] default arguments to sbcl, again
diff --git a/configure.ac b/configure.ac index 073f3e96..0c66c8de 100644 --- a/configure.ac +++ b/configure.ac @@ -452,6 +452,8 @@ case $fricas_lisp_flavor in sbcl) fricas_quiet_flags='--noinform --noprint --no-sysinit --no-userinit' fricas_eval_flags='--eval' + sbcl_memory_size=`echo "(princ (sb-ext:dynamic-space-size))" | $fricas_lisp $fricas_quiet_flags` + test $sbcl_memory_size -gt 2147483647 || fricas_quiet_flags="--dynamic-space-size 2048 $fricas_quiet_flags" ;; cmucl) fricas_quiet_flags='-batch -nositeinit -noinit' This should work. BTW, I should update configure with which autoconf, 2.71? Regarding "--disable-ldb --disable-debugger", they should not be used when dumping lisp image, instead they can be used when batch compiling lisp files. - Qian On 4/13/24 21:31, Waldek Hebisch wrote: On Fri, Apr 12, 2024 at 06:25:26PM +0800, Qian Yun wrote: We have discussed this before, namely the default dynamic-space-size is not enough during build. Recently there is a github issue hitting by this on freebsd. I get hit by this issue a few times as well, because the default "./configure" no longer works and sometimes I forget to use './configure --with-lisp="sbcl --dynamic-space-size 2048"'. I believe current situation is not friendly for new users, (and it's not documented explicitly in INSTALL), and cumbersome for developers. We should make "./configure" work by default. I think the following is a good solution: 1. Check the dynamic-space-size of the given SBCL. 2. If it is not enough, we append a proper argument to it. This has the advantage that the user can still specify a larger dynamic-space-size and will not be overwritten by our default value, which was the old plan in previous discussion. Yes, it would be good if you can do this. Also from the freebsd bug report [1], shall we use "--disable-ldb --disable-debugger" by default as well? "--disable-ldb" probably yes. "--disable-debugger" is somewhat debatable. Namely, when debugging build problems in typical cases most work is to get debugger prompt. Once in debugger, many problems can be easily diagnosed. Also, if you do "--disable-debugger" at sbcl command line, can you enable it later in running image? I hope so, just ask because I never used "--disable-debugger". -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/0e6a66f3-1a35-47b0-bc2b-54c2ae6de7a4%40gmail.com.
Re: [fricas-devel] default arguments to sbcl, again
On Fri, Apr 12, 2024 at 06:25:26PM +0800, Qian Yun wrote: > We have discussed this before, namely the default dynamic-space-size > is not enough during build. > > Recently there is a github issue hitting by this on freebsd. > > I get hit by this issue a few times as well, because the default > "./configure" no longer works and sometimes I forget to use > './configure --with-lisp="sbcl --dynamic-space-size 2048"'. > > I believe current situation is not friendly for new users, > (and it's not documented explicitly in INSTALL), and cumbersome > for developers. > > We should make "./configure" work by default. > > I think the following is a good solution: > > 1. Check the dynamic-space-size of the given SBCL. > 2. If it is not enough, we append a proper argument to it. > > This has the advantage that the user can still specify a > larger dynamic-space-size and will not be overwritten by > our default value, which was the old plan in previous discussion. Yes, it would be good if you can do this. > Also from the freebsd bug report [1], shall we use > "--disable-ldb --disable-debugger" by default as well? "--disable-ldb" probably yes. "--disable-debugger" is somewhat debatable. Namely, when debugging build problems in typical cases most work is to get debugger prompt. Once in debugger, many problems can be easily diagnosed. Also, if you do "--disable-debugger" at sbcl command line, can you enable it later in running image? I hope so, just ask because I never used "--disable-debugger". -- Waldek Hebisch -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/ZhqJSys6ACJbFMqf%40fricas.org.
