>I cross compiled pike 8.0.498 on an amd64/linux system for use on an
>armv5tejl/linux system, and I m getting a strange error when starting it up:
>
>-:1: Type mismatch for callback function `[]=:
>-:1: Expected: scope(0,function(zero, (0=zero) : 0)).
>-:1: Got : function(int(128..-129),
Windows?
> I'm trying to figure out what the "void|int share" argument of
> Protocols.HTTP.Server.SSLPort()->create() really does. The
> documentation is less than helpful:
>
> > Parameter share
> >
> > If true, the connection will be shared if possible. See
> > Stdio.Port.bind for more information
>
>
> Stephen R. van den Berg wrote:
[...]
> Incidentally the docs for Autodoc seem to be incomplete; it fails to
> document @global ?
It's probably intentional; from Tools.AutoDoc.DocParser:
parseError("The @global keyword is obsolete. "
"Use @belongs predef::
>As a by-product of creating the Val.Timestamp (et al) types, I encountered
>numerous bugs and uncertainties in mktime() and System.TM() with regard
>to timezone handling/abuse.
>
>AFAICS, all these bugs have now been dealt with and fixed in 8.1.
>Any objections if I port these fixes from 8.1 to
>I think adding a new module would be easier. The JSON5 format is very
>permissive and would add many new rules (e.g. identifiers as object
>keys). I think a new parser (maybe copied from the existing JSON parser)
>would make it easier to adjust the grammar.
>I am not sure if you like ragel, but I
>On Sat, Dec 23, 2017 at 12:42 AM, Henrik Grubbström (Lysator) @ Pike
>(-) developers forum <10...@lyskom.lysator.liu.se> wrote:
>> Note that AFAIK destruct() in some GTK2 classes is a public function
>> (and thus part of the API).
>
>Do you mean destroy? It got renamed in the big _destruct
Note that AFAIK destruct() in some GTK2 classes is a public function
(and thus part of the API).
>Henrik Grubbstr?m (Lysator) @ Pike (-) developers forum wrote:
>>>I propose ripping out the call_out() method, so that callouts are ran
>>>directly in all cases except for the timeout-case.
>
>>Not a good idea:
>
>> * Unlimited stack use.
>
>I re
>>b. Where did the docs go? I can't seem to find them anywhere.
>
>AFAIK, I copied the current docs to __builtin.Nettle.Hash.SCRAM; it
>should show up at the intended places due to inherits, but I haven't
>tried building the docs in a while.
I just tried building the docs, and SCRAM showed up
>Two things:
>a. I see a remark about making that cache-mapping weak. What would that
> help (it contains strings only)?
If the mapping were to be weak, any strings only referenced from the
mapping would be freed whenever the mapping hashtable needs to grow.
>b. Where did the docs go? I
>And speaking of Concurrent, since MutexKey is destructed immediately
>when it goes out of scope, I think many of the key=0 can be removed
>from the code.
I've not looked at this specific case, but you have to be careful with
tail calls and mutex keys, as it is possible for tail call
>Pike v8.0 release 509 running Hilfe v3.5 (Incremental Pike Frontend)
>> Crypto.SHA256.HMAC("a")("b");
>(1) Result:
>"\b\336""2\231""1\342\225h7v\252\232CR\233\320\262u(m\363\26\3\0\304\233\244\350A\203""0\23"
>> Crypto.SHA256.HMAC("a",32)("b");
>(2) Result:
>Branch: rosuav/hilfe-improvements
>
>Two smallish improvements to Hilfe. One is suppressing "quit" and
>"exit", per the TODO at the top; fairly simple but I'd like someone
>else's eyeballing to make sure it's going about things the right way.
Do you handle the case where the user ends the
The patch looks good to me.
> Why do we have two sets of unaligned memory access functions?
Good question.
> Both have exactly the same implementation, except for the obvious
> descripancy in the list of arches getting special treatment.
> Shouldn't we consolidate this to just one copy of the implementation,
> and a
>Since the object storage is zeroed, not calling the initializer is
>typically fine, but this initializer does
>
> SET_SVAL(THIS->lval[0], T_INT, PIKE_T_FREE, integer, 0);
>
>where PIKE_T_FREE isn't 0, so enabling this code would make a
>difference.
Looks like a bug.
Hi Arne.
>I would like to merge the branch arne/new_buffer. It implements
>and uses the new dynamic buffer implementation I have talked about on
>the conference. The main idea behind this implementation is to help the
>compiler generate better code in spite of the aliasing rules of C. In
>Pike 8.1 make doc results in:
>
>/var/src/roxen/81pike/lib/modules/Arg.pmod:623:Too many arguments to unknown
>function (expected 1 arguments).
>/var/src/roxen/81pike/lib/modules/Arg.pmod:623:Got : void | int.
Oops, seems I forgot to commit some typing fixes for the F_MAGIC_*
nodes. Does it
>I noticed that Debian package builds of both 7.8(.866-7) and 8.0(.358-1) have
>recently started failing on the powerpc architecture.
>
>https://buildd.debian.org/status/fetch.php?pkg=pike8.0=powerpc=8.0.358-1=1479857970
>
>/«PKGBUILDDIR»/build/linux-3.16.0-4-powerpc64-ppc/pike -DNOT_INSTALLED
I suspect that this behavior is another aspect of [bug 7723], and was
fixed with commit 57ebfa38c979402b6bdca02ab6593e570d963342:
| commit 57ebfa38c979402b6bdca02ab6593e570d963342
| Author: Henrik Grubbström (Grubba)
| Date: Fri Jun 17 17:34:34 2016 +0200
|
|
>Here are two:
>(gdb) bt
>#0 0x0055d953 in advance () at ./src/peep.c:910
>#1 asm_opt () at ./src/peep.c:1081
>#2 assemble (store_linenumbers=store_linenumbers@entry=1) at ./src/peep.c:441
[...]
>(gdb) cont
>Continuing.
>^C
>Program received signal SIGINT, Interrupt.
>0x0048a738
>hi,
Hi Martin.
>i just discovered that nettle in fedora/centos does not include the
>secp192 and 224 curves.
>this is not discovered during build, but instead just leads to a runtime
>error. (the installation succeeds despite the error)
That's a bit exotic, and would indicate that the
>Pike 8.0.294 beta/release candidate:
>
>https://pike.lysator.liu.se/pub/pike/beta/8.0.294/Pike-v8.0.294.tar.gz
A corresponding ebuild has now been added to the Gentoo Pike overlay.
Hi Arne.
>The documentation of the new f_hash() claims that its return value is
>architecture independent. SipHash reads its input as a series of
>little-endian 64 integers, which means that the hash value depends on
>endianess for wide strings.
Right, I forgot about that case.
>I have at some
>When preparing a new Debian package of 8.0.240, I noticed that some
>documentation disappeared, apparently due to the following. The error
>in Nettle.Sign is already corrected but not the others, AFAICT, and
>I'm guessing that @exp should be @expr. Patch below, which I home
>someone can apply
The precompiler should now work with a Parser.Pike that is more than 3
days old again...
>It seems that the 8.1 branch has been rewritten to work with
>Stdio.Buffer, which is good. But from what I can tell, SSL.File does
>not support that API. Is there a reason for this?
It's on the todo list for the SSL module.
> I recetly started getting this build error, which I have absoluteley
> no clue what to do about:
>
>
>[ ]
>
>
> In file included from
> /Users/ponost/src/pike-git/devel/src/post_modules/Nettle/cipher.cmod:16:
> /Users/ponost/src/pike-git/devel/src/module_support.h:47:1: warning: target
> does
>On Thu, Apr 28, 2016 at 1:40 AM, Henrik Grubbström (Lysator) @ Pike
>(-) developers forum <10...@lyskom.lysator.liu.se> wrote:
>>>Consider this usage:
>>>
>>>object stdout = Stdio.File();
>>>Process.create_process(cmd, (["stdout": stdout->pipe()]);
>>>
>>>Is this at risk of failing, where
>What exactly does PROP_IPC (fd_INTERPROCESSABLE in the source) permit?
Whether sending the file as eg stdin or stdout is supported by the OS.
The only platform where this flag is relevant is on WIN32, where
notably sockets don't have the flag set (cf src/fdlib.h).
>Consider this usage:
>
>I'm thinking about getting around to integrating Pike with the tzdata
>package in Debian, which I was talking about at the conference some
>years ago, but I need to check and understand some things first.
>
>If I understand correctly, the Calendar module knows about daylight
>savings time and
>The PIKE_MAPPING_KEYPAIR_LOOP is not enabled by default. Is this work
>in progress? Is it worth keeping?
Yes. It would allow for a more static traversal order for mappings.
With inlined byte code disassembly (and some braces and newlines):
>int main()
>{
=== 3 1c entry
=== 3 30 function start
=== 3 30 mark_at(0)
=== 3 54 pop to mark
=== 3 90 fill_stack(4,0)
=== 3 b6 init_frame(0,4)
>array(int) positions=({4,3});
=== 4 be constant(0)
>Incidentally, the reason I couldn't check in these fixes earlier is
>because they contained a rather oblivious bug.
>The code in question ran Thread.Thread(0) at some point (by accident), but
>the error message that this generates is devoid of any stacktrace or
>source code location information.
It seems something got broken in Stdio.UDP recently.
We are investigating...
/grubba
>So I thought I would fix that Gmp.mpz(3)%2 has the type mixed. My
>first quick and dirty fix was to update the type of `% a bit by adding
>
> tFunc(tObjIs_GMP_MPZ tInt, tObjIs_GMP_MPZ)
>
>Yes, it ignores the int%mpz and mpz%mpz permutations, but it's a
>start. Not the right one apparently, as I
>>>Warning: Wong return type.
>>>Warning: Expected: { mpz = object(implements _static_modules.Gmp()->mpz) }.
>>>Warning: Got : object(is _static_modules.Gmp()->mpz) | mixed.
>>>
>>>Why is there a "| mixed" in this type?
>>
>>You probably need to add a tIfNot (or modifying the existing) for the
The three new complaints about Assign instead of compare in
pike_tokenizer looks very, very intentional, but I'll leave it up to
whoever introduced them (grubba?) to mark them as false positives.
I believe that I acknowledged those three yesterday?
I think we should (1) issue a warning when Math.nan is used inside a
case and (2) make sure that switch (Math.nan) does not match any case.
I believe that the compiler should:
- Special case NaN if there's an explicit case-label with NaN.
- Map NaN to the default label (if any) otherwise.
Looks good to me. I was a bit worried about the first assignment, but
as both of the arrays are known to be of the same size, it is correct.
I have been working on refactoring the API of mega_apply and friends.
The purpose of that branch is to improve the performance of function
calls. More specifically, in some cases (e.g. f_map) frame allocation,
setup and deallocation could be avoided if the API permitted it. I will
soon push a
FYI: I've spent the last few days updating Pikefarm to current Xenofarm
and to monitoring git and the current branches of Pike.
It now seems to work again.
Caveat: The put binary in Xenofarm does not support redirects, so the
resulturl needs to be updated to pike.lysator.liu.se.
See branch: rosuav/curve_crypto_guard
Merged, thanks.
Looks like a quoting error.
I don't think I have any known major issues left to fix.
There was an issue with compiling Pike 8.x on machines without a pike.
Fixed in current Pike 8.x.
Add a test case to the testsuite?
Would anyone object to releasing 8.0 from 8.0.38?
Sounds good to me.
I don't think I have any known major issues left to fix.
Stephen R. van den Berg wrote:
Arne Goedeke wrote:
What kinds of things can safely be done from within destroy() and should
we document them?
Well, I for one, discovered that *if* any code in destroy() can throw
an exception, you must explicitly catch it, because if you don't the exception
will
Sounds ok to me.
So '%...x,str' would be equivalent to '%@...x,values(str)' for any
value of '...' I take it? I.e. you would use 'sprintf(%02x, Hej)'
as a shorthand for 'sprintf(%@02x, values(Hej))'?
That's what I assume, though I'd prefer
sprintf(%...x, Gmp.bignum(str, 256))
as equivalent in the first case
Martin Nilsson (Opera Mini - AFK!) @ Pike (-) developers forum wrote:
No, you could still call them. I kept them in because just because it
was compatibility with old data, not old code.
Ok, misjudgement from my side then.
Are there really permanently stored datastructures that need to be
I$-1òùll be in Link-Aöping visiting with Tobi and Company the weekend of
December 5-7. If anyone is interested, perhaps we could broaden the
scope to have more of a general Pike development meeting. Feel free
to drop a note (this list is probably a fine place) if anyone$-1òùs-A
On Wed, Oct 15, 2014 at 1:00 AM, Martin Nilsson (Opera Mini - AFK!) @
Pike (-) developers forum 10...@lyskom.lysator.liu.se wrote:
What do we need to make a 8.0 release this week, even if it is only
for Debian?
Possibly fix the docs building, which (last I checked) is failing in
8.1 and 8.0.
What do we need to make a 8.0 release this week, even if it is only
for Debian?
Left on my TODO-list for 8.0 is supporting the set_buffer_mode() API
in SSL.File (both for the SSL.File itself, and being set on the
supplied socket), but it can probably be left out in an initial release.
WIN32 eller WinSock.
Apologies for the non-minimality of the test-case, but this is a smidge weird.
The changes made in b2d25e Compiler: Improved soundness of the
op-assign optimizer seem to have broken this:
Thanks, that gave a hit at whet the problem was.
In theory, the three forms of element removal ought to be
Program.all_inherits has a FIXME asking for docs. But its definition
is a little odd: it returns a list of the given program's inherits,
and everything they inherit - but not recursively.
Not quite... Take a closer look at the loop -- It isn't a foreach()...
I could document it as such, but it
Test case:
int main() {write(Stdio.read_file(readfileboom.pike));}
Oddly enough, this works fine if run from a non-installed Pike:
rosuav@sikorsky:~/pike$ bin/pike readfileboom.pike
int main() {write(Stdio.read_file(readfileboom.pike));}
But installing it and using that causes a crash:
Thanks
Does it work with a Pike version that is a few hundred revisions old?
In that case bisect.
Fixed.
/home/nilsson/pike/src/testsuite.in:570: Test 140 (shift 2) failed.
1: mixed a() { return sprintf(%O, typeof(sprintf(%c, 1023))); ; }
2: mixed b() { return string(1023..1023); }
3:
o-a(): string
o-b(): string(1023..1023)
The above test is now fixed. It was due to
sprintf(%c, 1023)
While browsing some code (what else to do with your spare time ;-),
I find (not committed yet):
:-)
diff --git a/src/stralloc.c b/src/stralloc.c
index 7cd709c..6ca0a02 100644
--- a/src/stralloc.c
+++ b/src/stralloc.c
@@ -3221,10 +3221,8 @@ PMOD_EXPORT void free_string_builder(struct
Did you run
./run_autoconfig
?
It seems configure thinks that its a good idea to rerun the main
script in subdirectories that don't have their own configure scripts...
Isn't also TYPE_SUBTYPE wrong in the big endian case? It seems to me
that on a 64bit big endian machine both subtype and type would always
end up being 0.
Correct. Fixed last night.
like so?
#define TYPE_SUBTYPE(X,Y) (((Y)|((X)16)) (sizeof(ptrdiff_t) - 32))
No, sizeof(ptrdiff_t) needs to be
The large wide strings at startup are most likely the bytecode for the
master and related modules.
Does anyone have any feedback that would prevent me from announcing
7.8.856 as the new stable build? I was hoping to make some headway on
the windows ell problem I noticed, but if Peterâs build doesnât have
that problem, Iâm not too worried about having an answer to the
problem beforehand.
Hi Arne.
I get the above error with the following code. Shouldn't that work?
The parent use detector is overzealous...
arne
-
class A {
object D(string n) {
return C.D(this, E(n));
The problem here is that D has been marked as PROGRAM_USES_PARENT, but is
indexed directly
I would like to merge arne/slim into 8.0. It reduces text size with -O3
by about 100k. Most changes are probably harmless, however one disables
support for decoding programs with old style encoding and adds a
configure argument to turn it back on. I assume that part of the decoder
is not normally
There are some changes in the 7.8 git repository that would be nice to
have in the release to support the (soon to be released?) next version
of Nettle.
Hello all,
It's been a little over a week since I posted a new alpha
version, 7.8.852. Does anyone have any further feedback that needs to be
addressed before I announce it as a beta version on the primary list?
No objections here.
If I don't hear anything in the next day or so, I'll move it
Iâve uploaded a new 7.8 alpha, incorporating the latest changes for
bignum and clang. Please give it a spin. If I donât hear anything
Iâll promote it to beta and make an announcement on the main pike
list.
Pike-v7.8.852.tar.gz
Great! I've now updated the Gentoo Pike overlay
Good day,
Hi Bill.
I've prepared and uploaded a new Pike 7.8 alpha. This includes all of the
backports and has an up-to-date changelog. I've verified the build passes
tests* on OSX 10.9. Please have a look and let me know if there aren't any
problems. If all goes well, I'll upgrade it to beta
I believe using a here-document would be more portable. I've used
several UNIX(tm) systems that have lacked printf(1).
Hi Grubba,
Hi Bill.
Thanks for looking into this for me. I just tried to log in to
pike.lysator.liu.se and was prompted for a password. I did verify
that both ida and git accounts work properly with the keys I'm
using. Can you take a look to make sure things are okay?
After some
I've got a pike 7.8 alpha prepared, but I'm having trouble logging
in to pike.lysator.liu.se. Can someone with access verify my account
(bill) and key?
You should now have an account on pike.lysator.liu.se authenticated
with your usual set of ssh keys.
BTW: The ftp site is mounted from
This looks like a good opportunity to mention a few patches that I
haven't heard back regarding. They're mostly fairly trivial/simple. Is
there a better way to suggest patches than posting them to this list
or pike@roxen?
Either works.
Applied all of them except for the low_read_file() patch,
Consolidated notes from the Pike conference regarding Pike 8.
Pike 8 branch
-
[/] Update the CHANGES file.
[ ] Make the branch.
[...]
As far as I can see, there are no blockers left for making the branch.
If nobody protests, I'll make the branch later today.
And now the
However, nettle-2.1 patch failed on both 7.6.112 and 7.8.700 sources.
Thus, i stole src/post_modules/Nettle/cipher.c[mod] from 7.8.352, put it in
7.6.112, applied the nettle-2.1 patch from 7.8-stable/debian and ended up with:
[...]
So I took a look at the code now. Moveing from 2K (0x501) to XP
(0x501) APIs fixed the IPv6 problems.
Ok, I've bumped the value to 0x5ff in Pike 7.9.
I'm trying to boot up my old Windows build environment, and I'm getting a
new and exiting build error from master. Haven't looked at the code, but
will do that tomorrow night if no one has done it before then:
Making static: modules/_Stdio
Compiling modules/_Stdio/udp.c
Sounds good to me.
Before I try to work around that: What kinds of malloc misuses and
memory access errors will DEBUG_MALLOC actually detect, which valgrind
won't? Is it something we want to keep?
The main feature of DEBUG_MALLOC, is that it often is capable of
determining where the bug that generated the leak is
- Test that exported source and exported builds
compile/installs/work.
I believe that most of the decode_value-related problems have now been
fixed. Remaining to do is support for encoding of programs using
variants (the decoder should already be ok).
Implemented support for dumping of
Progress report:
The consensus was to attempt to push for a 8.0 release, given the
large amount of changes since 7.0, and to make that release with the
current feature set. The task list for that on a high level is
- Fork a 8.0 and next branch (9.0?) soonish (within a week?)
- Fix compilation
On Sat, 15 Jun 2013, Henrik Grubbström (Lysator) @ Pike (-) developers forum
wrote:
There are quite a few CritBit-related failures right now. Could they
be related to the renumbering of PIKE_T_*?
I think that is related to the stack renumbering. But I did not bisect
and also didnt
On torsdagen den 6 juni 2013, Magnus Holmgren wrote:
Excerpt from config.log:
configure:75019: checking if dynamic loading works
/home/holmgren/pike7.8-7.8.700/bin/smartlink gcc -c -g -g -O2
-fstack-protector --param=ssp-buffer-size=4 -Wformat
-Werror=format-security -O3 -pipe - ggdb3 -fPIC
FYI: The support for variant function (aka polymorphic overloading)
has now reached a semi-useable state in Pike 7.9:
| Pike v7.9 release 5 running Hilfe v3.5 (Incremental Pike Frontend)
| class foo {
| variant int bar(int a) { return a*2; }
| variant string bar(string s) { return FOO + s +
Magnus Holmgren wrote:
On onsdagen den 6 mars 2013, Neil McGovern wrote:
So, unfortunately the release team removed Pike entirely from the next stable
release of Debian, because I said that some of the bugs fixed by 7.8.700 were
rather serious. Can you help me elaborate on that?
was about twice as slow as the array_sscanf approach. Is that because
some special sscanf handling?
Probably due to the use of a lookup table.
BTW: Why don't you use wide_string_to_svalue_inumber() or STRTOL_PCHARP()?
Yes, but wasn't sscanf the fast code?
Anyway, if you want to try something exotic, you can try:
int low_hex2int(int c)
{
int q = c 0x40;
return (c + (q3) + (q6)) 0xf;
}
Oops, fixed.
I can't reproduce the problem on the main branch:
| Pike v7.9 release 5 running Hilfe v3.5 (Incremental Pike Frontend)
| -0x7fff;
| (1) Result: -2147483647
| -0x7fff - 1;
| (2) Result: -2147483648
| -0x7fff - 2;
| (3) Result: -2147483649
| Pike.get_runtime_info();
| (4) Result: ([
-NUMBER [(-$1a) 0] : NEG_NUMBER (-$1a)
-NEG_NUMBER [(-$1a) = 0] : NUMBER (-$1a)
+NUMBER [ !INT32_NEG_OVERFLOW($1a) (-$1a) 0] : NEG_NUMBER (-$1a)
+NEG_NUMBER [ !INT32_NEG_OVERFLOW($1a) (-$1a) = 0] : NUMBER (-$1a)
Hmm, the above changes hints that the problem is caused by a more
aggressive
Hi Bill.
Can someone grant me the ability to push branches? I've got a few
feature branches I'd like to get feedback on but I can't push them
because they have slashes in them (I assume it's preferred that I
not clutter the branch namespace unnecessarily.)
I believe that you should already
Hi.
I've now implemented the new rehash behaviour that we discussed
during the conference ~2 months ago. This means that weak mappings
that contain the only reference to a reference-counted object will
free the object when the hashtable is resized. Note that this only
takes care of the most
It should be up now.
The fixes look good to me.
I've prepared a beta release of 7.6 that contains all of the fixes
from the past few years, up through last week. Sadly, it doesn't
contain the clang patch that was just committed by mc, but I think
it wouldn't have made a big difference in overall clang
compatibility. I realize that many
I've just had to change my ssh keys due to a theft of equipment, and
I need to get my git access back. Can someone update my key(s)? I've
placed the new key(s) in my authorized_keys file on pike.ida.liu.se
Done.
101 - 200 of 431 matches
Mail list logo