Talk:GNU Multiple Precision Arithmetic Library

Is GMP required to build GCC or GCC is required to build GMP? Could someone please elaborate on that?

BOTH. Like any program written in C, GMP needs to be compiled with a C compiler. The GCC compiler will fill such a need. Later versions of GCC use GMP to convert calls to math functions where the argument is a constant into a constant. For example, when the compiler encounters sin(3.14), sqrt(25), etc., it will convert these into actual constants instead of function calls.

MPIR

I have added a link to MPIR an open source multiprecision integer (bignum) library forked from the GMP (GNU Multi Precision) and that has a windows friendly build system and that is compatible to GMP and it is used by many first class project (Sage_(mathematics_software). I have seen that any reference to MPIR library has been repeatedly deleted from this page, I hope that this time this link will remain. I sincerely hope that these deletions are not related to the fact that, according to many posts around the net, "MPIR was started as an angry fork of GMP". ALoopingIcon (talk) 22:37, 14 October 2010 (UTC)

Undid an anonymous, unmotivated attempt of removing the MPIR wiki reference. Please elaborate here before removing. ALoopingIcon (talk) 22:01, 1 November 2010 (UTC)

Again. Please elaborate here before removing. ALoopingIcon (talk) 08:50, 2 November 2010 (UTC)

This page is about the GNU project and specifically about GMP. MPIR is, if I understand it correctly, not GMP or part of the GNU project. Please write about MPIR in an MPIR page, where you might explain about your goals with it. Trying to motivate writing about MPIR on this page with a Sage reference is misleading and irrelevant, given that MPIR is a project organised withing Sage itself. Also, that Sage or some other project uses some package, does not motivate why that package should be dicussed in a page about GMP. —Preceding unsigned comment added by 130.237.222.220 (talk) 13:15, 2 November 2010 (UTC)

I sincerely do not understand why there is someone that want to remove a wiki reference to the MPIR library. MPIR is a LGPL library that forked from GMP to allow easier compilation for MS based compilers. I think that a reference here is quite well motivated, so I add it again. So, just to clarify, MPIR is a fork of GMP so a link in the GMP page is relevant. ALoopingIcon (talk) 13:55, 2 November 2010 (UTC)

Please do not add this link again without motivating that there should be a link to it. If the link is "well motivated", then let us hear the motivation! Else, please respect the subject of this article, which is GMP. Extension libraries (e.g. MPFR) and significant projects using GMP (such as Python, Ruby) are however relevant. The discussion of kaffe's previous use of GMP should probably be edited out, though. Note also that there is no links to the many other bignum libraries on this page; there is a separate page for that on Wikipedia. At that page, MPIR should absolutely be added (unless it is already there). —Preceding unsigned comment added by 130.237.222.220 (talk) 15:09, 2 November 2010 (UTC)

Re-inserted link. Link is clearly relevant: "MPIR a LGPL fork of GNU MP with fully compatible interface which (among other goals) aims to provide MSVC-based compilation system for Windows platforms." MPIR is a well established derivative work, and also it extends the library into another platforms. The fact that a project X is extended to platform Y is clearly a relevant thing to say about X. Therefore, MPIR should by all means be mentioned here.

The only reason I can think of why this link was repeatedly removed in the past is POV bias agaisnt MPIR. Please don't remove the link again. --Mtarini (talk) 16:47, 2 November 2010 (UTC)

Undid again. I kindly ask to the anonymous Stockholmer (all the anonymous censoring actions came from IP's located in Stockholm) to stop removing a the link to a legitimate LGPL fork. I would like to remember him/her that forks are quite relevant to the topic of the article, for example look at the FreeMind article: it has a section devoted to the forks it spawned (Freeplane and SciPlore_MindMapping). So, please, do not remove it again.

Moreover I would like remember him/her that if he/she is involved in the development of GMP, he/she should abstain from editing this page for plain lack of NPOV. (Useless disclaimer: I am not involved in any way in the development of MPIR). ALoopingIcon (talk) 21:56, 2 November 2010 (UTC)

Please stop this nonsense now. There is a page about MPIR here at Wikipedia, tagged for its poor quality. Please do productive work on that page instead of vandalising this page about a separate project. A list of bignum libraries might be another useful contribution. Making a POV list of bignum libraries here at the article about one specific bignum library is destructive, and moves away the from the focus of the article. The people reinserting the pro-MPIR lines should make serious contributions to the discussion, not ust say the linke is "well motivated". How do they motivate that other libraries that have taken GMP code are not mentioned, except POV? —Preceding unsigned comment added by 90.132.75.8 (talk) 06:47, 3 November 2010 (UTC)

I stop to revert your steady censoring actions to not incur in the WP:3RR. Again it seems that you do not accept the fact that legitimate forks are relevant to the article of an open source package. Raised the of NPOV on the noticeboard. ALoopingIcon (talk) 11:42, 3 November 2010 (UTC)

Agreed. This page is about GMP. GMP has been extended by MPIR so to work on a different platform (among other things). This is a piece of info *about GMP*, a relevant and even important one. MPIR is not "discussed" here, it is "mentioned", and the mention is totally relevant. Let's hope this senseless censorship by anonymous cease. --Mtarini (talk) 13:29, 3 November 2010 (UTC)

I suggest that ALoopingIcon and Mtarini is one and the same person, using two identities. MPIR is irrelevant in the context of GMP. MPIR does not work on any platform whatsoever not supported by GMP (not that this is relevant). One must keep the focus of a page here at Wikipedia, things that might seem slightly relevant from a POV perspective ("I like library X, so let's mention that in as many places as possible") cannot be added to a page about another library. Since if that principle was invoked, pages would end up very poorly focused. GNU TLS, GNU libc, which are truly relevant packages, have also "forked" GMP. And so have many other packages. The real reason why people add the link to MPIR is a MPIR bias. You like MPIR, therefore you want to scribble its name at a very public place. —Preceding unsigned comment added by 130.237.222.220 (talk) 14:54, 3 November 2010 (UTC)

This discussion is getting annoying, however since I'm pretty sure we don't want to resort to an RfC or even arbitration on this case we should find a common ground. On one hand the subject of forks is very touchy and sometimes it's hard to find sources to justify the inclusion of a unique pages for a new fork. On the other hand, by experience, an MPIR page would be merged with GMP. While I am not commenting on the inclusion or not of the MPIR link itself, it is clear that very common criticism of GMP are addressed by MPIR. A criticism section is quite common on the Wikipedia entries about open source software and GMP certainly could have one. In that case sourcing the said criticisms would require adding a link to some sources which would probably include an MPIR link. As such, I consider that either the link should be kept or that someone include a section addressing the common criticisms with relevant sources. BlanchardJ (talk) 16:33, 3 November 2010 (UTC)

As a start, could you please outline what common criticism of GMP is addressed by MPIR, or else any common criticism? I searched the GMP mailing lists, and there is very little criticism there. On the other hand, it seems that MPIR claims to uniquely support Solaris, Windows, Linux. While that can be taken as criticism of GMP, it is false, since GMP supports these environments, and many more. —Preceding unsigned comment added by 130.237.222.220 (talk) 17:12, 3 November 2010 (UTC)

Just a final, non polemic, bit of bare facts about MPIR and GMP. As a matter of fact GMP does not provide an easy support for MS compilers, particularly it seems not possible to compile GMP for 64bit architectures using ms compilers. According to many sites (and the FAQ of MPIR) these issues were the main reason of the forking from GMP. Just googling around you can find confirmations of the fact that MPIR offer better microsoft compiler support (for example on Stack Overflow [1], in various pages (avoiding mpir and sage pages obviously) [2], [3] [4] or even in the mailing list of GMP [5] where the only reasonable answer to a user asking how to compile GMP with MS VS 2008 was 'Consider using MPIR'.

To be honest, a couple of years ago, in the GMP mailing list a developer announced a future support of Microsoft Visual studio [6], but the date of the announcement (1st of April) and the overall tone of the post make me a bit suspect of the real will of include such a support... Ah. Just noted. The main developer of GMP, the one that according to various web sources is the one responsible of the absence of support of MS Compilers, is from Stockholm.

Just a final disclaimer, I am not definitely MTarini (neither BlanchardJ), and I am not involved at all with MPIR, I was involved in a project that required, at a given point, GMP compiled with VS2010 and I spent an afternoon of failed attempts before discovering MPIR. I added MPIR to the GMP page just to avoid similar headaches to other developers in a helpful spirit. Unfortunately I hit into an anonymous censor of Stockholm. ALoopingIcon (talk) 02:17, 4 November 2010 (UTC)

I see link was inserted again (by another user), and the usual anonymous user censored it again. I also see that, in the past, several other users added the link, and the Anonymous user censored it several times in the past. The arguments provided for removing the link (the one advocating "focus") do not hold any water (in which sense reporting a fork means losing focus on the subject). At this point, it is safe to assume that said anonymous user is probably among the authors or the main author of GMP, and is involved in some kind of "holy war" against the "angry fork". Needless to say, this holy war should be kept out of wikipedia, and authors should abstain from censoring pages about their creations. Any further censorship will be just considered for what it is, vandalism, and consistently reverted until stronger measures are adopted (hoping this won't be necessary).

Added the link back where it belongs. The suggestion by BlanchardJ is also very good, so added a criticism section too.

Unnecessary disclaimer: I am not remotely involved with the dev any of the libraries involved, nor I am a double identity of anybody. My an ALoopingIcon's account are both very old and we never posted on the same article before.--Mtarini (talk) 13:01, 4 November 2010 (UTC)

"Final facts"? "Non-polemic"??

[1] Incorrect; GMP builds under Windows and always did.

[2] You claim to be aoing MPIR pages, but this is a page from one of the most vocal MPIR people, judging from their own pages (he sometimes hides as "Cactus" there).

I didn't read [3] carefully, but yes, a 64-bit core2 specific build runs faster on a core2 on MPIR than a 32-bit pentium4 build. If it takes that sort of reasning to prove the MPIR\ is significant, then it is worse that I had expected.

I read [4] but I fail to see the point you're trying to make by linking to it.

What concerns [6], which developments of said Jarl Bong Rundtorn do you have in mind, and which make you call him a GMP developer?

I have no idea who "is responsible of the absence of support of MS compiler" (to you your wording). There are 6 billion people people in the world, aren't they all about equally responsible for not doing this work?

It would be nice to get back to the real question, though.

Could somebody make a start--here in the discussion section--to a section with (to use Blanchard's words) common criticisms against GMP. Not a sales pitch for MPIR, but real criticisms, and common for that matter.

When taking another group's work and starting to market it under your own name ("fork") the motives can be very different. Ego and attention-seeking plays an important role, often. Getting attention by trying to market the fork through the wikipedia paga about the original package is tempting, but I don't think it is correct to allow it.

I am undoing the latest sloppy edits. Here is the place to agree about what to do, refrain from controversial edits for a while, please!