On Friday, 29 August 2014 15:12:05 UTC+2, Jakob Kroeker wrote: > > Bill Hart:: >> ... There is a lack of documentation on what algorithms are implemented, >> what their complexities are, or references. Some projects are not >> threadsafe. Testing is lacking and quite a bit of stuff just doesn't work >> and never did. And there is a general lack of credit given to individual >> developers in Europe by their own projects. Most importantly there is a >> culture of not giving appropriate academic credit to individuals who have >> made significant contributions to writing mathematical software. >> >> I don't see that Sage has contributed to fixing a single one of these >> hard problems. >> > > I'm not sure if Singular/GAP did solve that problems either or has not > more problems. Even if in general it is good to have alternatives (to > Sage), in my opinion Singular+Gap is not the right way to do. >
I didn't know it had even had much of a chance to get going. I do know that it needs a lot more investment of time and money. > So I hope there will be a new CAS star at the sky which blasts the dated > Singular and GAP away. > I think that is the wrong approach. You are looking at it from a user's perspective, in that you'd like to have a nice Sage-like interface with all the functionality you require. But just writing a shiny new CAS doesn't give that to you. You still need the fast Groebner basis engine underneath, and the years of development that has gone into the multivariate arithmetic and to some extent the thousands of man-hours that have gone into libraries that use those things developed by mathematical researchers, who whilst working on those projects were the leading experts in their field. No amount of shiny CAS or even new funding, will give that to you all over again. Case in point. No less than 5 minutes ago, someone just showed me parallelised polynomial factorisation code that will be in Singular (hopefully) in the next release. It was developed as part of Singular's Factory. That represents years of investment by the DFG. That code, on a single core, was already faster than Magma. It is now much faster. Not a bad investment, I'd say! > > @referees: if you are reading this, just (let experts) evaluate the code > quality (readability, test coverage, correctness, maintainability, > extensibility,...) or the development process. You will not like the > outcome. Every couple of days I hit a bug in Singular and meanwhile I'm not > always officially file an issue in their bug tracking system because I was > asked to do so. > Unfortunately referees tend to not have time to examine large codebases in detail. They will however look at things such as past return on investment, the merits of the scientific/mathematical proposal, benefit to stakeholders, track record, how sound the proposal is from a technical perspective, their judgement of the likely outcomes,publication record, contribution to the progress of science. I'm pretty sure referees will acknowledge (even expect) that software systems contain bugs and that 20-30 year old systems will necessarily contain legacy code which hasn't yet been replaced or retired. So I think that even if referees did what you ask, and hired experts to evaluate the code, they'd find exactly what they expected to find (if competent). It would be very poor form to judge a scientific proposal on how perfect the code was, instead of on the scientific merits of the proposal and its importance and relevance to stakeholders. Singulars interpreter language has not even support of references (yes, on > a function call the data is (mostly) copied! ) and we are not talking about > data structures for trees or hash tables. > The Singular interpreter serves a completely different purpose than the Sage front-end. One of the things it has to keep doing for the foreseeable future is to support all the extant code written in the Singular language, for the original tasks Singular was designed to solve, and to support ongoing research work of the same kind (though the scope of the Singular project does seem to be growing lately). No new CAS can supply that demand. Nor should it! The current Singular interpreter may not provide you with all the modern language features, stability and performance you (or I) would like. But it it does provide important features, such as backwards compatibility with the huge legacy Singular codebase. To solve your problem you need new developers, paid with additional grant funding to work alongside the people maintaining the Singular interpreter, language, kernel and libraries, and definitely not for those people to stop what they are doing and work on something else! There's no point building a dog house if you don't feed the dog! If Singular were not maintained, all those thousands of man-hours of work would bitrot and become unavailable to the Open Source community, and it would/could not be replaced! The most valuable commodity of the Singular project is the existing code, and it needs direct and continued investment to maintain that functionality and to extend and improve it. (As things become truly relict, they can be cleaned away in the course of time. No one wants to maintain legacy code just for the kicks. But that happens naturally over time as projects evolve and as older features are no longer required, not because some users want something newer!) The problems you mention can all be solved, given time, money and innovation. The expertise certainly exists here in Europe. But it needs to be concentrated through ongoing funding. Obviously a larger collaboration is likely to bring the relevant expertise to bear where it is needed. Their groebner basis computation over intergers is buggy over years and > nobody noticed or cared ( well, I do, and now Adi Popescu and Anne > Fruehbis-Krueger are working on that issues ) > Great. > > Why is this important? Because otherwise you would be taking European >> money and using it to fund a project which originated in the US [...] This >> is crucial from the point of view of referees, in my opinion (again, please >> bear in mind this is my own personal opinion, and doesn't necessarily >> reflect the opinion of anyone else I have anything to do with). >> > > Unfortunately in the eyes of the referees some of this arguments (EU vs > US) could play a role, since what I'm heard from rumors was that one > argument of the Singular+GAP proposal was to have > an European alternative to the Sage project... > I was not aware that Sage played any part in the justification for proposals related to Singular+Gap. I'd be astonished if that were the case. But I have not read these proposals. I'd like to see a European software proposal based around Singular, Gap and Pari (perhaps others), regardless of the form it took. But it must address the (manifold and ongoing) needs of those projects directly! I will side with you in one respect. I don't think the Singular *language* and interpreter, in its current form is sufficient as the basis for the front end of such a project. In fact, I believe any such project should be language/project agnostic. There has been some talk of making the Gap language a standard. Some people seem to think it could be made so. I remain to be convinced of this. As you probably already know, I lean toward sophisticated, highly performant, modern programming languages. I don't think that is the aim or purpose of the Singular language, and I'm not sure about Gap (to my discredit, I know relatively little about it). Fortunately, I will not be making decisions about these issues on anyone's behalf! Bill. > > Am Freitag, 29. August 2014 14:30:16 UTC+2 schrieb Bill Hart: >> >> One other important point when interpreting my interjection (which again >> I stress is my own personal opinion), is that when mounting a campaign for >> a large grant, here in Europe or elsewhere, one must very clearly >> communicate what *need* is being addressed. If there isn't a clear need, >> you won't get money. >> >> If your project is perfect, and you go around telling everyone that it >> is, you will never be successful in getting the funding you desperately >> need. The reality here in Europe is that mathematical research projects of >> a computational nature struggle for their very existence. They are often >> run out of a single lab with one or two main developers and a few postdocs >> and PhD's if they are lucky. >> >> Some European mathematical software projects may not even be around >> tomorrow if they don't get appropriate funding. Not because they aren't >> worthy, or didn't have sufficient novelty, research value or smart people >> working on them or because they weren't feasible, but because everyone just >> assumed they would keep going on their own. They won't. >> >> These projects do not exist for the benefit of Sage. And in its current >> state, no matter how noble or well-intentioned or international Sage is, >> they can't! >> >> Sage does add value to those projects by widening the usership of those >> projects, by contributing bug reports and build patches, by bringing them >> publicity and in other ways. But one should never confuse this kind of >> support with sustained funding. In the same way, those project are >> benefiting Sage by being the best possible core components they can, given >> the heavy constraints they have on manpower, time and budget. But the end >> of those projects is not the enrichment of Sage, but the enrichment of >> their direct beneficiaries, which from the perspective of a grant >> application is the economy of the country who provided the grants, or the >> scientific enrichment of the union. >> >> To that end, we must, in my opinion, be very careful when applying for >> funds to "work on Sage". Who are the beneficiaries? How will they benefit? >> What is the strategy to achieve that goal? Is it sustainable, practical? >> What scientific merit does it have? How does it leverage the local >> expertise? What is its novelty? How well is it engineered? >> >> That's another 2c I'm owed for my personal opinion. >> >> Bill. >> >> On Friday, 29 August 2014 14:03:06 UTC+2, Bill Hart wrote: >>> >>> >>> >>> On Friday, 29 August 2014 13:17:40 UTC+2, Volker Braun wrote: >>>> >>>> First of all, it always saddens me when the ugly head of nationalism >>>> rears its head. I thought the time where we only support German science >>>> were over... >>>> >>> >>> You have misunderstood. When applying for German funding, the rules will >>> naturally state that the project must benefit the people paying for the >>> work, namely German companies and Mutter und Vater taxpayer. >>> >>> When applying for European funding, the rules will naturally state that >>> the funding must benefit the people paying for the work, namely the >>> European Union members. >>> >>> The idea that European funds should be used primarily to support an >>> international project *with no direct benefit to European projects* invoked >>> in the grant is patently a non-starter. That's just as bad, in my opinion, >>> as taking public funds to work on a closed source mathematical system! >>> >>> >>>> >>>> What sets Sage apart from GAP/Singular (and, I dare say: Flint) is the >>>> scale and the diversity of its contributors. >>>> >>> >>> No, what sets it apart is the number of contributors. Flint has had >>> contributors from all over the world. I would say from every continent >>> except Africa and Antarctica. >>> >>> We are talking about how to mount a campaign for European funding, not >>> about nationalising Open Source projects. And we are talking about the >>> maintainers and core developers of projects, not their volunteer >>> contributors. >>> >>> The reality is Singular is run out of Kaiserslautern, Germany, Pari out >>> of Bordeaux, France and Gap out of St. Andrews UK. They all have volunteer >>> contributors all over the place. But these are not paid employees, or at >>> the least volunteers paid by another University (to primarily do teaching >>> or research)! >>> >>> Flint is too small to be owned by a given university. The two core >>> developers currently aren't even at the same institution. It has received >>> EPSRC (UK support), DFG (German support), Austrian support and had a >>> developer at Harvard for a time. Even its maintainer (me) has been >>> supported from grants in two separate European countries! Flint has also >>> had salaries/stipends paid for from Google Summer of Code and from MSRI. >>> >>> Saying that it is a US (or European) project is just completely wrong. >>>> >>>> >>> It was started by William Stein at the University of Washington. A large >>> portion of the funding that built that project up came from grants of >>> William Stein and other funding he obtained, including from the NSF. He is >>> also in the process of trying to build a company in collaboration with the >>> University of Washington to make money to fund Sage development. >>> >>> The Sage Foundation is run through the University of Washington. If I >>> donate to the project, the money is handled by the University of Washington. >>> >>> There is no way that you can justify the assertion that Sage is not >>> primarily administered out of the US. And it has oodles of unpaid >>> developers all over the world. >>> >>> >>>> >>>> On Friday, August 29, 2014 11:46:14 AM UTC+1, Bill Hart wrote: >>>>> >>>>> This is all to say nothing of the glaring problems, such as the lack >>>>> of Windows 64 support >>>>> >>>> >>>> Wait, did you just do a 180 and say that we should drop everything to >>>> boost the market share of a failing north American software company? ;-) >>>> >>> >>> No. I never suggested that contributions of code should be made to >>> Microsoft. >>> >>> Bill. >>> >> -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
