Am 03.07.2015 um 00:03 schrieb Mathias Lang via Digitalmars-d:
> Congrats on the work done !
>
> 2015-07-02 22:39 GMT+02:00 Sönke Ludwig <digitalmars-d@puremagic.com
> <mailto:digitalmars-d@puremagic.com>>:
>
>
>     The second beta is out now. "dub init" now creates an SDL package
>     description by default (use --format=json to get a dub.json instead).
>
>
> Honestly, I am really not fan of implementing a feature AND making it a
> default in the same release.

It was a close decision, but the implementation is pretty simple and relatively well covered with tests, it shouldn't be an issue. So the only remaining issues would be possible compatibility issues for people not running the latest version (not *really* an issue) and the risk that the format, against all previous consideration, is deemed unsuited in the end (which would be really bad anyway).

What worries you the most?

Am 02.07.2015 um 22:45 schrieb rsw0x:
> On Wednesday, 17 June 2015 at 07:27:01 UTC, Sönke Ludwig wrote:
>> A new beta version out with support for the new SDL [1] based package
>> description format. If you are interested, please take some time to
>> test the implementation and to discuss any possible improvements to
>> the format specification before the final 0.9.24 release is made.
>>
>> The specification is currently still found on the wiki, but will be
>> moved to code.dlang.org together with the final release:
>> https://github.com/D-Programming-Language/dub/wiki/DEP1
>>
>> Download:
>> http://code.dlang.org/download
>>
>> Change log:
>> https://github.com/D-Programming-Language/dub/blob/master/CHANGELOG.md
>>
>> [1]: http://sdl.ikayzo.org/display/SDL/Home
>
> Have you considered referring to it as SDLang? SDL is difficult to
> google/already taken by a much more prominent project. I know SDL isn't
> your creation and I can't speak for anyone else but before it was
> proposed for dub I never heard of Simple Declarative Language.
>
> Googling "Simple Declarative Language" has less than 5,000 results so I
> guess I'm not alone.

I'd be very much in favor of that. We had a chat with the author about possible improvements, but I'm unsure if that topic has been addressed.

(Nick, do you remember by any chance where that discussion happened?)

On 07/03/2015 01:58 AM, Sönke Ludwig wrote:
> Am 02.07.2015 um 22:45 schrieb rsw0x:
>>
>> Have you considered referring to it as SDLang?

FWIW, the D project to read/write that format is named "SDLang-D" (for exactly the reasons you describe):

https://github.com/Abscissa/SDLang-D


> I'd be very much in favor of that. We had a chat with the author about
> possible improvements, but I'm unsure if that topic has been addressed.
>
> (Nick, do you remember by any chance where that discussion happened?)

It was all in a ticket (a few tickets?) for SDLang-D (and maybe DUB) on github, plus some direct email between marler8997 and Daniel (the original SDL guy).

I think this was the main one, not sure if there were others:
https://github.com/Abscissa/SDLang-D/issues/9

2015-07-03 7:06 GMT+02:00 Sönke Ludwig <digitalmars-d@puremagic.com>:

>
> It was a close decision, but the implementation is pretty simple and relatively well covered with tests, it shouldn't be an issue. So the only remaining issues would be possible compatibility issues for people not running the latest version (not *really* an issue) and the risk that the format, against all previous consideration, is deemed unsuited in the end (which would be really bad anyway).
>
> What worries you the most?
>

I'm just not comfortable making a format basically nobody knows, which was
introduced less than a month ago as "an additive change", which isn't as
well documented and indexed as the json one is (
http://code.dlang.org/package-format) the default.
In addition the likeliness that few bugs will arise doesn't mean that none
will, and I'd take a used-over-the-years solution over something just
implemented any day.
`dub init` is likely to be the first thing a beginner will want to do as
soon as (s)he learns about code.dlang.org, and event if it's still possible
to use json, it's not well documented / known yet, so I don't see a clear
return on investment on this change, while I can clearly see a net loss.

Obviously I'm not saying the format is not suitable, but I'd like to give it more time to mature.

On Friday, 3 July 2015 at 14:36:24 UTC, Nick Sabalausky wrote:
> On 07/03/2015 01:58 AM, Sönke Ludwig wrote:
>> Am 02.07.2015 um 22:45 schrieb rsw0x:
>>>
>>> Have you considered referring to it as SDLang?
>
> FWIW, the D project to read/write that format is named "SDLang-D" (for exactly the reasons you describe):
>
> https://github.com/Abscissa/SDLang-D
>
>
>> I'd be very much in favor of that. We had a chat with the author about
>> possible improvements, but I'm unsure if that topic has been addressed.
>>
>> (Nick, do you remember by any chance where that discussion happened?)
>
> It was all in a ticket (a few tickets?) for SDLang-D (and maybe DUB) on github, plus some direct email between marler8997 and Daniel (the original SDL guy).
>
> I think this was the main one, not sure if there were others:
> https://github.com/Abscissa/SDLang-D/issues/9

Since SDLang is a fairly obscure format I think we should mirror the docs somewhere. I was working on a project using SDLang-D and the spec website was down which was extremely frustrating because I couldn't find a mirror anywhere. SDLang-D's docs are pretty good so I was able to sort of work through it during the downtime but having the actual official docs somewhere extra would be good.

On 07/03/2015 03:16 PM, Brad Anderson wrote:
>
> Since SDLang is a fairly obscure format I think we should mirror the
> docs somewhere. I was working on a project using SDLang-D and the spec
> website was down which was extremely frustrating because I couldn't find
> a mirror anywhere. SDLang-D's docs are pretty good so I was able to sort
> of work through it during the downtime but having the actual official
> docs somewhere extra would be good.

That's a good idea. I'll put up a mirror of it as soon as a get a chance. Ping me on github (Abscissa) if I forget.

Am 03.07.2015 um 21:05 schrieb Mathias Lang via Digitalmars-d:
> 2015-07-03 7:06 GMT+02:00 Sönke Ludwig <digitalmars-d@puremagic.com
> <mailto:digitalmars-d@puremagic.com>>:
>
>
>     It was a close decision, but the implementation is pretty simple and
>     relatively well covered with tests, it shouldn't be an issue. So the
>     only remaining issues would be possible compatibility issues for
>     people not running the latest version (not *really* an issue) and
>     the risk that the format, against all previous consideration, is
>     deemed unsuited in the end (which would be really bad anyway).
>
>     What worries you the most?
>
>
> I'm just not comfortable making a format basically nobody knows, which
> was introduced less than a month ago as "an additive change", which
> isn't as well documented and indexed as the json one is
> (http://code.dlang.org/package-format) the default.
> In addition the likeliness that few bugs will arise doesn't mean that
> none will, and I'd take a used-over-the-years solution over something
> just implemented any day.
> `dub init` is likely to be the first thing a beginner will want to do as
> soon as (s)he learns about code.dlang.org <http://code.dlang.org>, and
> event if it's still possible to use json, it's not well documented /
> known yet, so I don't see a clear return on investment on this change,
> while I can clearly see a net loss.
>
> Obviously I'm not saying the format is not suitable, but I'd like to
> give it more time to mature.

It is heavily modeled after the JSON format, so it's not like there are a lot of new and unproven ideas. There may certainly be some things that could be improved, but those changes will need to happen in a backwards-compatible way no matter if default or not.

The documentation has been updated:
http://code.dlang.org/package-format?lang=sdl

Am 03.07.2015 um 16:36 schrieb Nick Sabalausky:
>
> It was all in a ticket (a few tickets?) for SDLang-D (and maybe DUB) on
> github, plus some direct email between marler8997 and Daniel (the
> original SDL guy).
>
> I think this was the main one, not sure if there were others:
> https://github.com/Abscissa/SDLang-D/issues/9
>

Thanks, found it by a cross reference:
https://github.com/D-Programming-Language/dub/pull/392#issuecomment-52092590b

Seems like the name hasn't been discussed at the time.

I can't understand why dub are trying to build vibed in app where it's does not implicitly linked:

app.d:
==============
import std.stdio;

void main()
{
	writeln("Edit source/app.d to start your project.");
}
=================
dub.json:
=================
{
	"name": "App",
	"description": "A minimal D application.",
	"copyright": "Me",
	"authors": ["Suliman],
	"dependencies": {
	"ddbc": "~>0.2.24"
	}
}
================
error:
================
C:\Users\di\AppData\Roaming\dub\packages\vibe-d-0.7.23\source\vibe\appmain.d(28): Error: static assert  "Error: -version=VibeDefaultMain is required to use vibe.d's default main(). Or use -version=VibeCustomMain to use your own main() instead. Please update your build scripts."
===============

Removing vibed from packages solved problem