Ok, I'm working on bit rot repair in my rational library.  Does anyone
know of a way, in light of Bug 3659
(http://d.puremagic.com/issues/show_bug.cgi?id=3659) to allow testing
two structs for equality when the r.h.s. is an anonymous rvalue or not
exactly the same type?  I refuse to commit a library where something
this basic doesn't work (especially since I'd have to make things
lvalues for no good reason in a zillion places in my unit tests), but it
looks like 3659 makes this impossible.

Also, is there any good reason why BigInt doesn't overload cast operators?  Should I fix this?  IMHO cast(long) someBigInt should work especially since BigInt already has a .toLong() method.

On 8/25/2010 10:58 PM, Andrei Alexandrescu wrote:
> This is a very interesting discussion, with great points. Here's what I think - in brief, my experience time and again has been that stuff that has a strong champion behind it succeeds, and stuff that doesn't, doesn't. In light of that, if Benjamin and/or David are enthused about pushing their libraries into Phobos, and furthermore maintaining and enhancing them, and further-furthermore contributing in new ways to Phobos, then I think it's not crucial that dimensional analysis (and possibly rational numbers) are of narrow utility.
>
> One less positive example is std.json - after adding it to Phobos, Jeremie didn't hang around to maintain it, bring its style on par with the rest of Phobos etc. Right now it's a library in need of a champion. Not to mention the likes of std.xml :o).
>
> So, subject to the rest of the team being approving too, I'd say let's give dimensional analysis (and rational numbers too if David wants) a fair shot at inclusion in Phobos. Be prepared for ruthlessness though :o).
>
>
> Andrei
>
> On 8/25/10 19:37 PDT, David Simcha wrote:
>> One comment I want to make is that, if we're going to include this, I have a rational number library on Scrapple that I've been meaning to put up for review for a while. I noticed that you roll your own somewhat ad-hoc rational numbers for the units type and don't really expose them in any more general way. IMHO if both rational and the units lib make the cut for Phobos, they should be integrated with each other. I'll clean up my rational library and do a commit, since it's succumbed to some degree of bit rot. It's at http://dsource.org/projects/scrapple/browser/trunk/rational/rational.d.
>>
>> On 8/25/2010 10:09 AM, Benjamin Shropshire wrote:
>>> I have offered up a library that supports statically encoding units in the type system so as to prevent unit errors (adding distance and time) and to enforce correct conversions all around.
>>>
>>> http://d.puremagic.com/issues/show_bug.cgi?id=3725
>>>
>>> I'm looking for comments: What's holding it back from inclusion? What
>>> would need to be improved? The API? Better comments? (I haven't tested
>>> it recently so; make it build again?)
>>> _______________________________________________
>>> phobos mailing list
>>> phobos at puremagic.com
>>> http://lists.puremagic.com/mailman/listinfo/phobos
>>>
>>
>> _______________________________________________
>> phobos mailing list
>> phobos at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/phobos
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>

AFAIK there is no solution. I had the same issue with cursors in dcollections.

Sent from my iPhone

On Aug 25, 2010, at 11:15 PM, David Simcha <dsimcha at gmail.com> wrote:

> Ok, I'm working on bit rot repair in my rational library.  Does anyone know of a way, in light of Bug 3659 (http://d.puremagic.com/issues/show_bug.cgi?id=3659) to allow testing two structs for equality when the r.h.s. is an anonymous rvalue or not exactly the same type?  I refuse to commit a library where something this basic doesn't work (especially since I'd have to make things lvalues for no good reason in a zillion places in my unit tests), but it looks like 3659 makes this impossible.
> 
> Also, is there any good reason why BigInt doesn't overload cast operators?  Should I fix this?  IMHO cast(long) someBigInt should work especially since BigInt already has a .toLong() method.
> 
> On 8/25/2010 10:58 PM, Andrei Alexandrescu wrote:
>> This is a very interesting discussion, with great points. Here's what I think - in brief, my experience time and again has been that stuff that has a strong champion behind it succeeds, and stuff that doesn't, doesn't. In light of that, if Benjamin and/or David are enthused about pushing their libraries into Phobos, and furthermore maintaining and enhancing them, and further-furthermore contributing in new ways to Phobos, then I think it's not crucial that dimensional analysis (and possibly rational numbers) are of narrow utility.
>> 
>> One less positive example is std.json - after adding it to Phobos, Jeremie didn't hang around to maintain it, bring its style on par with the rest of Phobos etc. Right now it's a library in need of a champion. Not to mention the likes of std.xml :o).
>> 
>> So, subject to the rest of the team being approving too, I'd say let's give dimensional analysis (and rational numbers too if David wants) a fair shot at inclusion in Phobos. Be prepared for ruthlessness though :o).
>> 
>> 
>> Andrei
>> 
>> On 8/25/10 19:37 PDT, David Simcha wrote:
>>> One comment I want to make is that, if we're going to include this, I have a rational number library on Scrapple that I've been meaning to put up for review for a while. I noticed that you roll your own somewhat ad-hoc rational numbers for the units type and don't really expose them in any more general way. IMHO if both rational and the units lib make the cut for Phobos, they should be integrated with each other. I'll clean up my rational library and do a commit, since it's succumbed to some degree of bit rot. It's at http://dsource.org/projects/scrapple/browser/trunk/rational/rational.d.
>>> 
>>> On 8/25/2010 10:09 AM, Benjamin Shropshire wrote:
>>>> I have offered up a library that supports statically encoding units in the type system so as to prevent unit errors (adding distance and time) and to enforce correct conversions all around.
>>>> 
>>>> http://d.puremagic.com/issues/show_bug.cgi?id=3725
>>>> 
>>>> I'm looking for comments: What's holding it back from inclusion? What
>>>> would need to be improved? The API? Better comments? (I haven't tested
>>>> it recently so; make it build again?)
>>>> _______________________________________________
>>>> phobos mailing list
>>>> phobos at puremagic.com
>>>> http://lists.puremagic.com/mailman/listinfo/phobos
>>>> 
>>> 
>>> _______________________________________________
>>> phobos mailing list
>>> phobos at puremagic.com
>>> http://lists.puremagic.com/mailman/listinfo/phobos
>> _______________________________________________
>> phobos mailing list
>> phobos at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/phobos
>> 
> 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

On 26 August 2010 04:58, Andrei Alexandrescu <andrei at erdani.com> wrote:
> This is a very interesting discussion, with great points. Here's what I think - in brief, my experience time and again has been that stuff that has a strong champion behind it succeeds, and stuff that doesn't, doesn't. In light of that, if Benjamin and/or David are enthused about pushing their libraries into Phobos, and furthermore maintaining and enhancing them, and further-furthermore contributing in new ways to Phobos, then I think it's not crucial that dimensional analysis (and possibly rational numbers) are of narrow utility.
>
> One less positive example is std.json - after adding it to Phobos, Jeremie didn't hang around to maintain it, bring its style on par with the rest of Phobos etc. Right now it's a library in need of a champion. Not to mention the likes of std.xml :o).

How about std.openrj in Phobos1! That's a situation we desperately want to avoid.

> So, subject to the rest of the team being approving too, I'd say let's give dimensional analysis (and rational numbers too if David wants) a fair shot at inclusion in Phobos. Be prepared for ruthlessness though :o).

That seems a fair argument.

Now that the meta topic is out of the way, anyone have comments on my code? :)

I think we should move forward with this. In the interest of exposing this to a broader audience, I'll post the link and my initial comments to the digitalmars.d newsgroup.

Happy New Year Phobosians! (Phobosites? Phobozens? Phobolings? Phoban?)


Andrei

On 8/25/10 9:09 AM, Benjamin Shropshire wrote:
> I have offered up a library that supports statically encoding units in the type system so as to prevent unit errors (adding distance and time) and to enforce correct conversions all around.
>
> http://d.puremagic.com/issues/show_bug.cgi?id=3725
>
> I'm looking for comments: What's holding it back from inclusion? What
> would need to be improved? The API? Better comments? (I haven't tested
> it recently so; make it build again?)
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos