Pike-devel January 2018

pike-devel@lists.lysator.liu.se

8 participants
19 discussions

Sql: big_typed_query()
by Stephen R. van den Berg 25 Jan '18

25 Jan '18

Why was it decided that if big_typed_query() is not supported by an sql driver, that we want to throw an error instead of transparently falling back to big_query()? I normally would say, the transition to big_typed_query() is, by definition, a fuzzy one, because depending on the database (driver), one cannot possibly have a one-to-one typed equivalence for all available database types. So it is inevitable that some values will be typed, and others will still be strings. And with that in mind, you might as well fall back to big_query() completely if there are no typed results. If there are no objections, I'll alter the code accordingly (at least in 8.1). -- Stephen.

1 0

The GET_VAL macro and C99
by ceder (-) Per Cederqvist ＠ Pike (-) developers forum 24 Jan '18

24 Jan '18

The GET_VAL macro in src/builtin.cmod is invoked like this: > GET_VAL (true) > GET_VAL (false) > GET_VAL (null) If you happen to include <stdbool.h>, the macro expansion might result in these functions: get_val_1 get_val_0 get_val_null This is because true is defined as 1 and false as 0. This isn't a big issue for two reasons: - src/builtin.cmod does not actually invoke <stdbool.h> - as far as I can see, nothing actually uses get_val_true and get_val_false (but I may be overlooking something) This could be fixed by ## and # directly in the GET_VAL macro, instead of going via PIKE_CONCAT and TOSTR. Of course, that breaks on older compilers that don't support ## and #; I don't know if the Pike community still cares about them. One could also argue that "if it ain't broke, don't fix it". That is a valid point of view. But the code is a bit fragile; sooner or later somebody might add <stdbool.h> without that it affects the GET_VAL expansion.

2 2

//
by Martin Nilsson (Coppermist) ＠ Pike (-) developers forum 20 Jan '18

20 Jan '18

Given the amounts of C++ comments in the code, is there actually a platform we care about that can't handle them?

3 3

Protocols.HTTP.Server.SSLPort()->create([...], void|int share)
by ceder (-) Per Cederqvist ＠ Pike (-) developers forum 18 Jan '18

18 Jan '18

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 (Source: <http://pike.lysator.liu.se/generated/manual/modref/ex/predef_3A_3A/Protocol…>) But the Stdio.Port.bind documentation does not mention "share". Looking at the source code I see that my suspicion is correct: "share" corresponds to the "reuse_port" argument. I think "share" should be renamed "reuse_port", so that it is easier to read the documentation. Any objections? If not, I'll commit such a change to the 8.1 branch tomorrow.

3 2

Fun with strict_types
by Tobias S. Josefowitz 17 Jan '18

17 Jan '18

Hi all, I've had a little run in with strict types, and I fail to convince myself that certain aspects of its behaviour have merit. Consider $ cat test.pike #pragma strict_types void main() { array(array(mixed|string)) to_be_mapping = ({ ({ "foo", 1 }), ({ "bar", "xxx" }), }); mapping(string:int|string) m; #ifdef WORKAROUND mixed mix(mixed m) { return m; } #else # define mix(X) (X) #endif m = [mapping(string:int|string)]mix(mkmapping(to_be_mapping[*][0], to_be_mapping[*][1])); } $ ~/local/bin/pike test.pike test.pike:17: Warning: An expression of type mapping(mixed:int | string) cannot be assigned to a variable of type mapping(string:int | string). $ ~/local/bin/pike -DWORKAROUND test.pike (no output indeed) Am I correct to assume that silently - somehow - not accepting my softcast in at least some regards on something like mapping(mixed:int|string) while accepting it on mixed might not be intended behaviour? Tobi

1 0

strftime() and strptime() indirect via System.TM?
by Stephen R. van den Berg 15 Jan '18

15 Jan '18

While adding strftime() and strptime() methods to Val.Timestamp and Val.Date, I couldn't help but noticing that in order to access both libc functions I need to go through System.TM. Wouldn't it be preferable to create direct native builtins for those two functions instead? (to put them in the same boat with mktime/gmtime etc?) -- Stephen.

1 0

AmigaOS... and Pike 8.1
by Stephen R. van den Berg 12 Jan '18

12 Jan '18

I just ripped out AmigaOS from Pike 8.1. Martin warned me (after I'd already ripped it out, of course) that maybe Marcus is still using it? So, is anyone still using AmigaOS in Pike 8.1? If yes, I'll revert the patch. If no, or if we stop supporting at 8.1, I'll leave it as is. -- Stephen.

3 8

What is the difference between _encode() and _serialize() ?
by Stephen R. van den Berg 12 Jan '18

12 Jan '18

What is the difference between _encode() and _serialize() ? -- Stephen.

1 0

GTK/GDK on OSX
by H William Wellliver III 07 Jan '18

07 Jan '18

I’m running into some build problems on OSX. It seems that the GTK module assumes that if a system is NT that gdkwin.h is used and otherwise gdkx.h is used. Unfortunately, it seems that on some builds of GTK/GDK for OS X, neither of those is correct and that gdkquartz.h should be used. I’m guessing that some configure tests are needed to determine which of the 3 files is actually present. However a quick look at configure.in leaves me unsure how best to include that. Anyone with more expertise in the GTK module’s inner workings care to weigh in? Bill

4 9

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

Pike-devel January 2018