gmane.comp.lang.pike.devel

Pike 7.8 Release (last call)

H. William Welliver III — 2012-05-22T03:16:43

Does anyone have any objections to 7.8.680 becoming the next official release of Pike? Do note that machine code is non-functional on (at least more modern instances of) i386. If I don't hear any objections in 24 hours, I will attempt to update the website, etc, with the new information.

Best,

Bill

strict type regression

2012-05-16T15:30:02

I've started getting

/home/nilsson/pike/lib/modules/String.pmod/module.pmod:98: Warning: An
expression of type string | int | !mixed cannot be assigned to a
variable of type int.

after the changes in there. It should be possible to work out that

  int sz = min(< at >map(strs, sizeof));

is always the correct type.

pike_frame->fun = FUNCTION_BUILTIN

Arne Goedeke — 2012-05-05T22:25:17

In INIT, EXIT and some gc callbacks the fun in Pike_fp is set to
FUNCTION_BUILTIN. there are some places in the code (e.g.
ADT.List()->_get_iterator) where ->fun is used in INHERIT_FROM_INT et
al, which segfaults (except with PIKE_DEBUG). I am not sure where to
fix this. Should the *_FROM_INT macros check for FUNCTION_BUILTIN
or is this better to be handled in the individual INIT/EXIT methods ?

does anyone object to making ADT.Stack and ADT.Queue inherit ADT.List?

arne

Visualization of inherit-graphs

2012-05-02T18:45:02

Hi.

I've gotten several requests for adding inherit-graphs to the Pike
documentation. I now have a proof of concept that renders the graphs
as a <pre>-formatted block similar to the corresponding feature in
javadoc (but supporting multiple-inherits). Some examples:

| Class SSL.https
| Inherit graph
| 
|     predef::ADT.Queue
|      |  
|      |  predef::SSL.context
|      |   |  
|      |   |  predef::Stdio.Port
|      |   |   |  
|      +---+---+--predef::SSL.sslport
|                  |  
|                  +--predef::SSL.https

| Class Parser.XML.Tree.SimpleRootNode
| Inherit graph
| 
|     predef::Parser.XML.Tree.AbstractSimpleNode
|      |  
|      |  predef::Parser.XML.Tree.VirtualNode
|      |   |  
|      |   |  predef::Parser.XML.Tree.XMLParser
|      |   |   |  
|      +---+------predef::Parser.XML.Tree.SimpleNode
|              |   |  
|              +---+--predef::Parser.XML.Tree.SimpleRootNode

| Class GTK2.FileChooserDialog
| Inherit graph
| 
|     predef::GTK2.Object
|      |  
|      +--pre

Metafunctions

2012-04-27T10:45:01

I've got a bit of a problem with the `->() lfun.  To illustrate,
consider the following class:

---8<--
class Base
{
  mixed `->(string x)
  {
    return
      (x == "add_one" && lambda(int x) { return x+1; }) ||
      (x == "add_two" && lambda(int x) { return x+2; });
  }
}
---8<---

I now have a class which allows me to do Base()->add_one(7) to get 8,
and Base()->add_two(7) to get 9.  So far so good.

Now, let's try to specialize this class:

---8<---
class Sub
{
  inherit Base;
  string add_two(int x) { return x+"two"; }
}
---8<---

FAIL!  Sub()->add_two(7) still returns 9.  And what's more, even if I
try to add something with a name that doesn't exist in Base, I still
can't access it outside of Sub.

Is there a way around this?  I tried using `::->() as a fallback in
`->(), but it didn't have the desired effect.

Pike 7.8 beta: feedback?

Bill Welliver — 2012-04-23T23:01:44

So, 7.8.680 has been out and about for a week, and despite the fact 
that I checked a fix in after I made that build, I'm inclined to say .680 
should become an official release. Does anyone have any feedback? Go? No 
go?

Bill

Segfault

2012-04-23T15:25:02

I get this crash when trying to build pike with dmalloc.

#### Making dynamic: post_modules/GTK2
make[2]: Entering directory `/home/nilsson/pike/dmalloc/post_modules/GTK2'
Makefile:576: warning: overriding commands for target `depend'
Makefile:332: warning: ignoring old commands for target `depend'
/home/nilsson/pike/dmalloc/pike -DNOT_INSTALLED -DPRECOMPILED_SEARCH_MORE -m/home/nilsson/pike/dmalloc/master.pike  /home/nilsson/pike/src/post_modules/GTK2/build_pgtk.pike --source='/home/nilsson/pike/src/post_modules/GTK2/source/' \
                   /home/nilsson/pike/src/post_modules/GTK2/output/few.pike
make[2]: *** [compile1] Segmentation fault
make[2]: Leaving directory `/home/nilsson/pike/dmalloc/post_modules/GTK2'
make[1]: *** [GTK2] Error 1
make[1]: Leaving directory `/home/nilsson/pike/dmalloc/post_modules'
make: *** [post_module_objects] Error 1

And it looks like it is in the startup things bail out.

nilsson< at >riaa:~/pike/dmalloc$ gdb pike
GNU gdb (Ubuntu/Linaro 7.3-0ubuntu2) 7.3-2011.08
Copyright (C)

Pike 7.8 Beta

H. William Welliver III — 2012-04-18T05:51:50

I've prepared a new 7.8 test release for consideration. Please feel free to download and test.

http://pike.ida.liu.se/download/pub/pike/beta/7.8.680/

I am not aware of any show stoppers in this release; if none are found in the next week, this will become the new stable 7.8 release.

As always, please feel free to get in touch with any comments or concerns.

Best,

Bill
bill-NuqkLFvhzZFg9hUCZPvPmw< at >public.gmane.org

map/filter, mapping index/values

Lance Dillon — 2012-04-12T19:20:24

I'm playing with map and filter and stuff, trying to do a functional way of removing an index from a map, like this:

({ 

  ([ "name": "user1",
        "val": "test",
        "random": "other"

  )],
  ([ "name": "user2",
        "val": "another test",
        "random": "something else"
  ]),
})

The array and mapping will be in that kind of format.  name is unique among all the map entries, and all the keys exist with all the maps.  I want to change it to something like this:

([ "user1": ([ "val": test", "random": "other" ]),
    "user2": ([ "val": "another test", "random": "something else" ])
])

I can get the keys for the new map with array[*]["name"].

But removing that key from the maps and leaving the others is giving me a problem.  Filter doesn't quite work because it only gives the value, and I'm not testing on the value, but the key.

A foreach would work, but I'm wondering if I can do something like:

mkmapping(a[*]["name"],map(a,lambda(mapping m) { filter(......

7.8 release note assistance sought

Bill Welliver — 2012-04-11T15:25:05

I've been working through the check-in comments and have about 3/4 of them 
distilled into meaningful release notes, however I'm getting to the point 
where I need a little assistance weeding out noise and making sense of 
what's left.

I put together a list of the entries that I haven't added to the release 
notes yet, and I'd like to give everyone an opportunity to weigh in. The 
list is at the following url:

http://wiki.gotpike.org/ReleaseToDo7-8-664

If you've got something to say about anything left, feel free to put 
together a sentence or two describing the change. You can either 
send them to me, or just add them to the bottom of the wiki 
page. Also, if you know the entry isn't worth writing about, 
feel free to remove it from the list. Otherwise, I won't put too much more 
effort into adding the rest of these.

Best,

Bill

mime type update

Bill Welliver — 2012-04-05T18:51:33

Any objections to adding a few newish popular mime types to the MIME 
extension database? Specifically, I was thinking about the ODT types, but 
would be glad to add others if someone cares to send them my way.

Bill

new block allocator

Arne Goedeke — 2012-04-02T21:16:48

We pushed an experimental branch which uses a new block allocator we 
have been developing. This one uses a different algorithm than the one 
we briefly mentioned on the conference. The new one uses a hashtable for 
page lookup instead of radix trees.

The branch can be found under arne/block_alloc, a GJAlloc-bundle is 
available at http://fortranland.com/GJAlloc-0.08.tar.gz or from 
https://github.com/arneg/GJAlloc via git. The branch currently contains 
alot of obsolete history, which we would squash into one commit in case 
of inclusion.

Standalone performance comparisons are available at 
https://github.com/arneg/GJAlloc/wiki/Performance-comparison. The 
speedup is quite noticeable when compiling with --cleanup-on-exit, 
otherwise for most use cases the differences should be unnoticable. 
However, when having many things around that do not get freed linearly, 
the speedup can be dramatic. A rather extreme example is

  time pike -e 'array a = map(allocate(1000000, 5), random_string); for (int i = 0; i <

gobject-introspection

2012-04-01T10:35:01

Hi guys.

I was feeling bored so I started work on a Pike binding for GLib:s
object introspection repository (aka "gir").

It's still horribly incomplete, but it can already handle simple stuff
like this:

---8<---
import GI.repository;

void main()
{
  Notify.init("notifytest");
  Notify.Notification.new("Hello world", "This is an example",
  "dialog-information")->show();
}
---8<---

The branch is available as "marcus/gobject-introspection" on the git
server if you want to check it out or (*gasp*) contribute.  :-)

String handling optimisation

Stephen R. van den Berg — 2012-03-29T08:22:13

Does anyone know how often in the code we actually depend on the
fact that the same string will be at the same address in memory?

Because I'm contemplating an optimisation which would involve making
the string duplication avoidance opportunistic instead of mandatory.

I.e. something along the lines of all strings shorter than stringmin will
always be optimised to a single reference, and all strings above that *might*
have more than one reference, but not necessarily do (i.e. they're not
fully hashed all the time, to avoid the overhead of rehashing large strings
repeatedly when juggling around lots of strings).

All the places that depend on same string = same address would need to be
patched.  Also, to determine stringmin, some profiling of existing apps
would be interesting.  Is that statistic available for say Roxen, to know
the distribution of string length and reference count in a running
application?

Compiling Pike 7.9 crashes?

Stephen R. van den Berg — 2012-03-28T14:05:50

#### Making dynamic: post_modules/Bz2
#### Making dynamic: post_modules/COM
#### Making dynamic: post_modules/CritBit
/var/src/roxen/rpike/build/linux-3.1.9-vs2.3.2.5-dirty-x86_64/precompile.sh precompile.pike --api=4 >"/var/src/roxen/rpike/src/post_modules/CritBit/inttree.c" "/var/src/roxen/rpike/src/post_modules/CritBit/inttree.cmod" || { rm "/var/src/roxen/rpike/src/post_modules/CritBit/inttree.c"; exit 1; }
precompile: pike /var/src/roxen/rpike/bin/precompile.pike --api=4 /var/src/roxen/rpike/src/post_modules/CritBit/inttree.cmod (method=QQQQQQ)
Unsupported API version: 4 (max: 3)
precompile: /usr/bin/pike /var/src/roxen/rpike/bin/precompile.pike --api=4 /var/src/roxen/rpike/src/post_modules/CritBit/inttree.cmod (method=QQQQQQ)
Unsupported API version: 4 (max: 3)
precompile: /var/src/roxen/rpike/build/linux-3.1.9-vs2.3.2.5-dirty-x86_64/pike -DNOT_INSTALLED -DPRECOMPILED_SEARCH_MORE -m/var/src/roxen/rpike/build/linux-3.1.9-vs2.3.2.5-dirty-x86_64/master.pike  /var/src/roxen/rpike/bin/precompile.pike --api=

pgsql fixes and other things

Stephen R. van den Berg — 2012-03-28T12:25:09

I just pulled the latest pike version from git.
It clearly has been a while, since I've actively participated.
I apologise for my extended leave, it was due to some real-life issues
that got in the way of things.  That is not to say that I didn't stop using
Pike or Roxen; quite the opposite, I just didn't have time to keep up on
the mailinglists.  I got snippets of updates through Marc though, so I wasn't
completely ignorant.

Anyway, the good news is: I'm back.
I just glanced through the mailinglist archives (in my mailbox) and it is
quite a bundle to skip through (not started on that yet).
I first checked the state of the pgsql module.  I see some minor updates/fixes.
I have one small and one very important fix waiting here which should get in
ASAP (which I have been hunting down for the past three years, but did not,
until about 2 weeks ago, result in a SEGV; which finally made it possible
to zoom in on it; which, in effect, was the final nudge that I needed to get
back on track with the mailinglist here)

abi and building with bundles

Bill Welliver — 2012-03-27T17:06:55

I noticed a few problems that exist when building Pike using bundles. I'm 
not very familiar with how that all works, so I thought I'd report them 
here in the hope that someone more comfortable with making changes to this 
can take a closer look:

1. If you specify an ABI that isn't the "default" for a given system 
using --with-abi, the ABI specific compiler flags don't seem to be passed 
on to any subsequent bundles. Thus, you get all kinds of problems with 
mismatched abis when trying to link. A temporary workaround seems to be to 
define CFLAGS and pass them in for the whole build process.

2. A problem specific to Nettle seems to be that there's a small c++ test 
program that gets built as part of the Nettle bundle. When building for a 
non-default ABI, this test program can't link against the built library, 
causing the whole bundle to fail. A temporary workaround seems to be to 
define CXXFLAGS and pass these flags (such as -m32) as part of the overall 
build process.

Bill

Pike 7.8 Alpha

H. William Welliver III — 2012-03-22T03:50:07

I've prepared a new 7.8 test release for consideration. Please feel free to download and test.

http://pike.ida.liu.se/download/pub/pike/alpha/7.8.664/

The following known problems exist:

- Change log is not complete (feel free to contribute)
- A compilation problem exists under OSX 10.6/7: pike_threadlib includes mach*.h, which causes an error.

Please feel free to get in touch with any comments or concerns.

Bill
bill-NuqkLFvhzZFg9hUCZPvPmw< at >public.gmane.org

need permissions help on pelix

H. William Welliver III — 2012-03-21T02:38:22

I'm trying to create some new release folders and unfortunately the permissions seem to be incorrect (I'm in the pike group):

$ ls -l /pike/data/ftp/pub/pike/
total 20
-rw-r--r--   1 nilsson  pike         804 Sep 14  2004 README
drwxrwsr-x  33 jhs      pike        1024 Mar 19 22:12 all
drwxr-sr-x  12 peter    pike         512 Jan  3  2011 alpha
drwxrwsr-x  14 peter    pike         512 Apr 16  2011 beta
drwxr-sr-x   2 jhs      pike         512 Aug 16  2002 contrib
drwxr-sr-x   2 jhs      pike         512 Aug  2  2002 docs
drwxr-sr-x   2 mast     pike        1024 Mar 19 22:06 latest-stable
drwxr-sr-x   2 jhs      pike         512 May 12  2001 libs
drwxr-sr-x   2 jhs      pike         512 Aug  2  2002 man
drwxr-sr-x   4 jhs      pike         512 Aug  2  2002 patches

Can someone help out with that?

Thanks,

Bill

paging Peter Bortas

H. William Welliver III — 2012-02-16T04:00:45

I don't have Peter's email address, so hopefully he's monitoring this…

I'd like to get working on a 7.8 release but don't really have any idea what's involved, aside from change log entries (which I'm working on, slowly). Anything you might have for me to go on in that department (even if it's a rough outline) would be greatly appreciated!

Bill

compilation handler for multi-tenant applications: update

Bill Welliver — 2012-02-08T23:19:45

Hi all,

A few months ago, started a conversation about what I'll call multi-tenant 
applications: that is, a single pike process running multiple applications 
(perhaps including copies of a given application) with isolated program 
and module spaces. The idea is to provide a similar capability to that 
provided by Java's ClassLoader API.

That is, I'd like to be able to do the following:

                                pike
                +-----------------|-----------------+
                |                                   |
          thread a1...an                     thread b1 ... bn
     compilation handler a                compilation handler b
  loads classes for app1 instance 1  loads classes for app1 instance
       from location a,b                    from location a,c (perhaps)
      uses module foo.bar                  uses module foo.bar

            foo.bar              !=             foo.bar


            though it would perhaps be acceptable if

      object_program(foo.bar)    ==