Hi folks,
Here's a patch that drops the "extension-information" procedure.
As Felix points out on the chicken-5-new-egg-install-notes page ("drop
it completely? we can't locate extensions, just eggs"), this procedure
doesn't really make sense anymore, and it would need to be modified to
do what
LGTM. Here's a sign-off.
Thanks,
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Hi there,
Looks good to me, applied. This is a nice simplification, I think. I had
to tweak the patch slightly to account for Peter's recent "platform"
patch, but it wasn't too different.
Cheers,
Evan
___
Chicken-hackers mailing list
Hi there,
Excellent, applied.
Thanks Felix!
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Hi Peter,
Good stuff, pushed!
On 2017-04-22 22:29, Peter Bex wrote:
> While working on the patch, I noticed that provide and provided? are not
> really related to feature identifiers as such, and more to code loading,
Yep, that's absolutely right (discussed on IRC, but figured I should
mention
Hi Mario,
These look good, all of them have been applied. Thanks!
On 2017-04-23 20:18, Mario Domenech Goulart wrote:
> They are probably trivial. If they look ok and you want me to push
> directly, just drop me a note.
They are, but in reviewing them I learned a few things so I don't
Hi Peter,
That seems fine to me, applied.
I don't think it's worth trying to fix this in the 4.x line, so I've
closed the ticket as fixed.
Cheers,
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
Hi Peter,
You're quite right, pushed.
Evan
signature.asc
Description: PGP signature
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Hi Kristian,
This was a good catch, thanks for the patch!
Pushed to master and chicken-5.
Best regards,
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
This moves the dotfile and symlink-related behaviour checks for
`delete-directory` into posix-tests.scm and out of the test scripts
themselves, and removes a potentially dangerous operation that would
symlink to a system directory when deleting files.
---
tests/posix-tests.scm | 36
I don't totally follow, but I trust the rest of this thread's judgement.
Pushed.
Evan
signature.asc
Description: PGP signature
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Ugh.
Pushed.
Evan
signature.asc
Description: PGP signature
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
This makes sense to me. It's also nice to move those ad-hoc `eq?` checks
in the `##core#set!` handling into a common location.
On 2017-04-09 16:23, Peter Bex wrote:
> PS: The ##core#set! hunk is so big because it fixes a bad indentation.
> I went through the trouble of replacining [] by () as
The C_process_sleep macro, introduced in 0493e29, depends on
for access to Sleep() on Windows, so this converts it into a procedure
in runtime.c where we already include that header. It also renames the
procedure to reflect the fact that it's an inline call.
---
chicken.h| 7 +--
On 2017-03-01 22:59, Evan Hanson wrote:
> the other option is to make C_process_sleep an extern'd function in
> runtime.c (rather than a macro), which would avoid the need to move
> any include statements around. So, if anyone feels that it would be
> better not to change these includ
We can probably remove it, but for now I think we should leave it alone
until the dust settles on the new toolchain and we can see whether it's
really a false positive, or helpful for the common use cases.
Evan
___
Chicken-hackers mailing list
Good catch, added.
Thanks Jörg.
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
On 2017-03-25 11:34, Peter Bex wrote:
> On Sat, Mar 25, 2017 at 12:11:38AM +0100, lemonboy wrote:
> >
> > here's a small patch that makes `string->list' tail recursive, we gain
> > a small speed up and one less word allocated for every iteration.
> >
>
> Looks good to me, and should indeed go to
Hi Peter,
LGTM. All three have been pushed.
Cheers,
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Hey folks,
Pushed, with the minor change to the conditional we discussed on IRC.
Thanks Lemon!
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Like the other modules in posix.scm, this just reexports identifiers
from the "chicken" and "chicken.posix" modules.
Also, add `executable-pathname` to chicken.import.scm, to which it had
never been added (so it was available implicitly at the toplevel but
totally inaccessible from within a
On 2017-03-10 8:35, Peter Bex wrote:
> It looks like you forgot about the header of the bignum vector.
Ah, you're totally right.
That makes sense, thanks for the clarification.
Pushed!
Evan
signature.asc
Description: PGP signature
___
On 2017-03-10 11:53, Evan Hanson wrote:
> We can change the behaviour on EINTR in a later commit, if we decide
> it should indeed change.
And by "EINTR" I meant, of course, "EWOULDBLOCK".
Evan
signature.asc
Description: PGP signature
_
On 2017-03-10 3:24, Kooda wrote:
> On Fri, 10 Mar 2017 10:25:36 +1300
> Evan Hanson <ev...@foldling.org> wrote:
> > I think `call-with-exec-args` should use `check-environment-list`
>
> I wanted to ask about that in my initial email but forgot about it
> before sendin
Hi folks,
On 2017-03-04 17:44, Peter Bex wrote:
> On Mon, Feb 06, 2017 at 05:32:20PM +0100, Jörg F. Wittenberger wrote:
> > This was at least the behavior the code did before.
> >
> > But that would be an incompatible change wrt. the current behavior.
> > Hence I did not mix this with the
Hi Kooda,
Very nice, I like this change.
On 2017-03-02 0:40, Kooda wrote:
> diff --git a/posix-common.scm b/posix-common.scm
> index f8fe27fa..f3d444c9 100644
> --- a/posix-common.scm
> +++ b/posix-common.scm
> @@ -758,6 +768,9 @@ EOF
>
> ;; Envlist is never converted, so we always
Hi all,
On 2017-03-05 23:35, felix.winkelm...@bevuta.com wrote:
> IIRC, all the module changes are not documented, yet, is this correct?
>
> Would it be a good idea to start documenting these changes? Speaking only
> for myself, I completely lost my grasp on what functionality is provided by
>
On 2017-03-04 7:35, felix.winkelm...@bevuta.com wrote:
> Could you add the egg also to the chicken-5-eggs repo?
Yep, done.
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
On 2017-03-02 10:34, felix.winkelm...@bevuta.com wrote:
> Is this for CHICKEN 5? If not, then we need a change request, as this is an
> incompatible change.
I think this, and in fact all changes that aren't strictly bug fixes for 4.12,
should go into chicken-5 only from this point on. I know
The `compile-file` and `compile-file-options` procedures have been moved
to an egg, so the utils unit is now empty and can be dropped entirely.
Also fix the types.db entries for `qs` and `system*`, which have been
moved to the chicken.process module.
---
NEWS
On 2017-03-02 10:36, felix.winkelm...@bevuta.com wrote:
> I can't recall whether we have settled on this, but having a "file"
> and a "files" module strikes me and somewhat confusing.
Right, it's my intention that we drop "files" once it (and posix) have
been refactored into "file" and other
Stubs EWOULDBLOCK on Windows so that the module only needs to be
declared once.
---
posix.scm | 43 +++
posixunix.scm | 44
posixwin.scm | 47 ---
3 files changed,
Also adds a stub for `signal/bus` on Windows so that the list of
signal/* values is the same on both platforms.
---
README| 1 +
chicken-install.scm | 1 +
defaults.make | 2 +-
distribution/manifest | 2 ++
posix.scm | 10 ++
posixwin.scm
Hi hackers,
Here's a small patchset that adds most of the modules on the roadmap
that replace the current posix module. It's a fairly small change since
it mostly just adds new module forms rather than properly refactoring
posix into separate units, reason being that I started trying to do
This module simply reexports the low-level file-related values from
posix (mostly having to do with file descriptors).
It also adds a stub for `file-control` and the various various fcntl/*
values on Windows so that the posix module's export list can be the same
across all platforms.
---
README
This is a placeholder for what will eventually be the (chicken file)
module. For now, it simply reexports values from the preexisting "files"
and "posix" units.
---
README| 1 +
chicken-install.scm | 1 +
defaults.make | 2 +-
distribution/manifest | 4 +++
eval.scm
---
README | 2 +-
chicken-bug.scm | 2 +-
chicken-install.scm | 4 ++--
chicken-status.scm | 2 +-
chicken-uninstall.scm| 2 +-
csi.scm | 2 +-
defaults.make| 2 +-
distribution/manifest| 8
eval.scm
This inverts the way the posix unit is organised by having one outer
file (posix.scm) that includes a platform-specific implementation file
(either posixunix.scm or posixwin.scm), depending on what feature flag
is passed.
---
distribution/manifest | 1 +
posix.scm | 46
---
README| 1 +
chicken-install.scm | 1 +
defaults.make | 3 ++-
distribution/manifest | 2 ++
posix.scm | 6 ++
rules.make| 3 +++
6 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/README b/README
index 623d868a..9d960704
This module reexports all process-related procedures from the posix
unit, as well as `qs` and `system*` from utils. The former procedure has
been tweaked to avoid a dependency on `string-intersperse` from the
data-structures unit, while the latter has been modified so as not to
"sprintf" its
When `system` is called with a string containing null bytes, the
resulting error should use "system" as its location, not
"get-environment-variables".
---
runtime.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/runtime.c b/runtime.c
index 9ccce654..f6ac6866 100644
---
As of 0493e29 we always need access to Sleep() on Windows because it's
used in the "C_process_sleep" macro, so this includes early
in chicken.h rather than in runtime.c. It also drops a redundant include
from setup-api.scm, which uses the same #ifdef conditional as a guard.
---
chicken.h | 7
Hi hackers,
Here's a small patch for chicken-5 that fixes a build issue on MinGW
introduced by, I think, 0493e29:
In file included from library.c:11:0:
library.c: In function 'f_26703':
chicken.h:1602:30: warning: implicit declaration of function 'Sleep'
[-Wimplicit-function-declaration]
Pushed. Thanks Peter, and thanks to Rob for the report.
Evan
signature.asc
Description: PGP signature
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
LGTM, Pushed.
Cheers,
Evan
signature.asc
Description: PGP signature
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Hi folks,
Some test results for the counting, inline below.
The short version is that everything works except 64-bit MinGW builds
(i.e. the "mingw" and "mingw-msys" platforms), which failed due to issue
#1344 (https://bugs.call-cc.org/ticket/1344).
Cheers,
Evan
>From 5fbefcebe8d1ce34c81986093399730d0941eeeb Mon Sep 17 00:00:00 2001
From: Evan Hanson <ev...@foldling.org>
Date: Thu, 19 Jan 2017 20:33:25 +1300
Subject: [PATCH] Fix spurious array bounds warnings on OpenBSD
This changes the C_block_item struct definition to always use a
C99-style flexible array for
On 2017-01-20 16:28, Kooda wrote:
> Looks fine, pushed!
Thanks Kooda!
> I’m curious though: how did you catch this?
http://cppcheck.sourceforge.net/
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
This changes the fallback definition of C_block_item used for non-C99
compilers to use a slight variation on the "struct hack", where instead
of using 1 for the size of its data field, we use the largest possible
size. This increases the value that the compiler uses for bound analysis
in order to
This reuses our aliases from chicken.h and silences a couple of linker
warnings on OpenBSD.
---
dbg-stub.c | 45 ++---
1 file changed, 22 insertions(+), 23 deletions(-)
diff --git a/dbg-stub.c b/dbg-stub.c
index b41ec346..beb61b5b 100644
--- a/dbg-stub.c
Hi Peter,
Good catch! Pushed.
Thanks much,
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Oy vey.
Pushed, with the include moved up in the file as discussed.
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Looks good, pushed to master.
Not pushed to chicken-5, as discussed.
Evan
signature.asc
Description: PGP signature
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
On 2017-01-16 21:31, Peter Bex wrote:
> I've performed a system update of my Cygwin install (which was at
> least a year old or so), and that seems to have fixed the problem.
Excellent, glad that's sorted it.
> Attached is a modified version of the patch which sets _XOPEN_SOURCE
> to 700, which
Hi guys,
Thanks for looking at the patches.
I've pushed a modified version of Peter's fix for time on Cygwin, since
that's clearly an improvement over relying on `command'. (We can't use
"true" in time's place, however, or the tests it's timing won't actually
be run since true totally ignores
This provides access to strptime(3) from , which is otherwise
undefined on at least Cygwin and possibly other platforms.
---
chicken.h | 4
1 file changed, 4 insertions(+)
diff --git a/chicken.h b/chicken.h
index 4ae93e09..dec79aa3 100644
--- a/chicken.h
+++ b/chicken.h
@@ -47,6 +47,10 @@
---
rules.make | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/rules.make b/rules.make
index 09fdd76f..ee93821e 100644
--- a/rules.make
+++ b/rules.make
@@ -372,7 +372,7 @@ install-bin: $(TARGETS) install-libs install-dev
$(prog)$(EXE)
Hi folks,
Here's a small patch that fixes a Makefile bug on Windows. This was
necessary when installing on MinGW-w64. I'm not sure why we hadn't run
into it before -- perhaps it's a difference between MinGW (the "normal"
one) and MinGW-w64?
Anyway, I think it should be applied to both master and
Hi Kooda,
Looks good, thanks for the fix. Applied.
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
---
tests/runtests.sh | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/tests/runtests.sh b/tests/runtests.sh
index 745ad1c..164c1f4 100755
--- a/tests/runtests.sh
+++ b/tests/runtests.sh
@@ -87,6 +87,7 @@ SETUP_PREFIX="${SETUP_PREFIX} -e (register-program \"csi\"
On 2016-12-11 18:00, Peter Bex wrote:
> I think it's counter-intuitive that module or functor definitions are
> expressions that can even yield a value.
Just wanted to mention that this is actually one of my favorite things
about our module system. It's occasionally handy to be able to isolate a
Thanks Peter, both have been applied. This stuff is hairy, but the
approach of using a second let for ordering (or rather, *not* ordering)
is nice. Thanks for looking into the issue.
And many thanks to Joo ChurlSoo as well, for uncovering and reporting
it. The test case is very much appreciated.
Hi all,
Good changes, applied.
Evan
signature.asc
Description: Digital signature
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Hi Peter,
Wow, thanks for doing this. It looks like it was quite a bit of work,
and I'm all for it. Pushed.
One thing that's now inaccessible but *might* be worth reexposing is the
binary version. For example, salmonella currently uses it to set up a
custom repository extension path, and I've
On 2016-09-29 20:02, Peter Bex wrote:
> As promised, here's the patch to drop C_BUCKET_TYPE in favor of using
> simple pairs in the symbol hash table bucket lists.
Pushed.
Nice work with all of this, Peter.
Evan
signature.asc
Description: Digital signature
Patches pushed, ticket closed. Nice find, Peter.
Evan
signature.asc
Description: Digital signature
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
that adds a
"_with_loc" variant instead, like you suggest.
Evan
From d70236672a5dfba4dbe1b2e0c15385c451822e1e Mon Sep 17 00:00:00 2001
From: Evan Hanson <ev...@foldling.org>
Date: Sat, 10 Sep 2016 22:53:36 +1200
Subject: [PATCH] Detect and signal error on stack overflow in `apply`
When a
LGTM, pushed.
Thanks,
Evan
signature.asc
Description: Digital signature
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
suming the performance
impact is negligible.
This patch applies atop the previous one in the thread, so if it looks
OK I'm happy to push that one first.
Cheers,
Evan
From e05f8eed98a1f46a4f922e9a77f89970701af6cc Mon Sep 17 00:00:00 2001
From: Evan Hanson <ev...@foldling.org>
Date: Thu, 8 Sep 2016 22
Hi folks,
Looks good, thanks for digging into this issue. I've pushed these. I
also restored the return value of process-spawn, which was changed by
the patch (unintentionally, I think) but must preserve the result of
spawnvp[e] in order to be useful.
Cheers,
Evan
signature.asc
Description:
Adds a new include form that searches for files relative to the
including file rather than the CWD. In all other ways it behaves like
the normal `include`.
Because `load-relative` and `include-relative` are so similar, this
change also moves some pathname handling code into a group of shared
Hi Peter,
Good improvements all, pushed.
Regarding your note about special-casing `vector-length`, I'm curious
why you think it wouldn't make sense? It seems worthwhile to me.
Evan
signature.asc
Description: Digital signature
___
Chicken-hackers
Hi folks,
Looks good, applied to chicken-5. It didn't apply cleanly to master; was
it meant to? Personally, I'm fine making this a 5-only change.
I also updated the mdocs for the "-module" and "-link" options, which
are different and new in 5, respectively. There might be some others
that need
On 2016-07-19 12:14, Timo Myyrä wrote:
> Yeah, those are my patches.
Ha, well, in that case thanks and nice work. It sounds like the other
hackers are generally open to this as long as it can be plugged into the
build with relatively little pain, so a self-contained patch for this
would be
Hi Timo,
On 2016-07-19 10:53, Timo Myyr� wrote:
> What do you think?
I'm very much in favor of a change in this direction.
Note that zmyrgel has already done most of the work to produce new
mandoc sources; see http://bugs.call-cc.org/ticket/1177#comment:1.
Cheers,
Evan
As pointed out by Peter on IRC, providing a hint about what kind of port
is expected when the wrong one is given is valuable from a
user-friendliness point of view, so this restores the error messages for
the typical input- and output-only cases, and uses the more general
wording only when the
On 2016-06-30 12:17, felix.winkelm...@bevuta.com wrote:
> I didn't compile or run this, but is it possible that "(test ..." is
> missing in the tests above?
Ah, you're correct of course. Those will have nooped because it's
compiled "-analyze-only". And in fact they were spurious since the
Makes sense, applied.
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
Convert the port direction and closed flags from booleans to bitmasks,
to allow for multidirectional ports.
---
NEWS | 2 ++
chicken.h| 6 +++--
library.scm | 74 ++--
ports.scm| 14 +-
Adds a new "refine" type form to the scrutinizer, which allows other
types to be "tagged" with extra information. Applies this approach to
the existing types for input- and output- ports, which are now expressed
as "(refine (input) port)" and "(refine (output) port)", respectively.
Improves the
This allows an input and output port to be combined into a single
bidirectional port.
---
NEWS | 3 +++
library.scm | 2 +-
manual/Unit ports| 9 +
ports.scm| 29 +
tests/port-tests.scm | 33
Hi hackers,
Here is a quite large patchset that, at its heart, adds support for
multidirectional ports by converting the simple boolean flag indicating
whether a port is for input or output to a bitmap. It also adds
`input-port-open?` and `output-port-open?` procedures for testing
whether a port
These test whether a port is open in a specific direction.
---
NEWS | 2 ++
chicken.h| 10 --
chicken.import.scm | 2 ++
library.scm | 19 +--
manual/Unit library | 12 +++-
tests/port-tests.scm | 19 +--
Looks great, applied. That inconsistency between "error" and
"error-output" always bugged me.
I agree that things like this should only be going into 5 at this point.
Thanks for the suggestion, Michael!
Cheers,
Evan
signature.asc
Description: Digital signature
Previously, ##sys#kill-other-threads wouldn't clear the ##sys#fd-list,
leading to errors when unrecognized file descriptors were encountered
during scheduling. Additionally, it would set slots in random heap
locations near ##sys#fd-list (potentially leading to heap corruption)
since it incorrectly
Makes `sleep` change its behaviour based on whether the scheduler is
loaded: when it is, the current thread is suspended (allowing other
threads to continue executing), and when it isn't the whole process is
suspended with sleep(3). Also adds a `process-sleep` procedure to the
posix unit as a way
Hello,
Here's another bit of work from the roadmap: it makes the behaviour of
`sleep` more intuitive in the presence or absence of threads. See the
commit message for details.
There's an accompanying patch for srfi-18 (attached) which I'm happy to
apply if this looks good.
Cheers,
Evan
Index:
These procedures have been moved to an egg.
---
Makefile.aix | 1 -
Makefile.android | 1 -
Makefile.bsd | 1 -
Makefile.cygwin | 1 -
Makefile.haiku| 1 -
Makefile.hurd | 1 -
Makefile.ios | 1 -
Makefile.linux| 1 -
Makefile.macosx | 1 -
---
README| 1 +
chicken-install.scm | 1 +
defaults.make | 4 +-
distribution/manifest | 2 +
manual/Unit posix | 1 +
posixunix.scm | 102 +++--
posixwin.scm | 103
---
README| 7 +-
c-platform.scm| 102 +++---
chicken-install.scm | 1 +
defaults.make | 2 +-
distribution/manifest | 2 +
lolevel.scm | 215 --
manual/Unit
Hi folks,
Here are some library reorganization patches for chicken-5.
The first creates a "chicken.memory" module per the "ICC minutes" on the
reorg page[1]. I only moved the procedures that were "Decided", so there
are still a bunch left in lolevel that need a home. Those are:
block-ref
LGTM, pushed.
Note that this change doesn't affect csi, since `repl` installs its own
error handler. That's currently using 2048 as a *per-argument* print
limit, via `##sys#repl-print-length-limit`. Might be worth modifying
that to match compiled behaviour.
Evan
signature.asc
Description: PGP
When a thread that's waiting on another with `thread-join!` is forced to
execute -- usually on an interrupt, when the primordial thread is
forcibly resumed via `##sys#force-primordial` -- it must return control
to the scheduler if it can't yet be unblocked. Otherwise, execution will
return to the
Now that "collapsable-literal?" constants are inserted into the
constants table as quoted values, there's no need to box them in a list
(which was done to ensure that "#f" isn't treated as a missing entry).
---
core.scm | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git
Hi Peter,
Good eye! Applied.
We can even go a bit further still and get rid of the boxing of values
in the constants table, by the same line of reasoning. Patch attached.
Cheers,
Evan
___
Chicken-hackers mailing list
Chicken-hackers@nongnu.org
Hi John,
On 2016-05-28 11:51, John Cowan wrote:
> It seems to me that it should have the highest priority, so that if a
> specific version of some library code is packaged with the code, it
> will be used in preference to any installed version.
It will be preferred to the CHICKEN_REPOSITORY
Hi folks,
Here's another patch for chicken-5. It makes "(include ...)" search for
files relative to the including file. Relative includes are given the
lowest search priority, i.e. they are only used when the included file
isn't found in the CWD or any of the directories specified with the
Allows the following program to work correctly:
(define-constant a 'frizzle)
(print a)
Previously, the second `a` would be replaced by an *unquoted* `abc`,
resulting in an undefined variable reference (or, if the constant value
were instead `(quote a)`, causing the compiler to enter an
One for chicken-5. See the commit message for an explanation.
A case *could* be made for the current behaviour as a way to do
compile-time variable substitution, but IMO `define-constant` is the
wrong way to support such a feature and as it stands it's impossible to
define a plain symbol as a
Fantastic stuff, thanks Kooda. Applied to master and chicken-5.
As a nice side effect, testing this patch uncovered some unrelated but
questionable behaviour that would be good to fix, too (#1286).
Cheers,
Evan
signature.asc
Description: PGP signature
401 - 500 of 714 matches
Mail list logo
