-
-
Notifications
You must be signed in to change notification settings - Fork 434
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Make lcalc compatible with the new PARI #11321
Comments
Reviewer: Volker Braun |
comment:1
Looks good to me. Are you making a full lcalc spkg at one point? |
comment:3
Replying to @vbraun:
Yes, when I have time. |
This comment has been minimized.
This comment has been minimized.
Dependencies: #11130 |
Upstream: Reported upstream. Little or no feedback. |
comment:7
New spkg ready for review, needs to be built with the new PARI spkg. |
Changed upstream from Reported upstream. Little or no feedback. to Reported upstream. Developers acknowledge bug. |
comment:8
On 2011-05-12 18:56, Michael Rubinstein wrote: Thanks for the patch. I should point out that I now have a google code page for lcalc: http://code.google.com/p/l-calc/ (lcalc was taken, so I went with l-calc). The page has my beta newer version of lcalc, L-1.3 I should point out that Lcommandline_elliptic.cc has become part of Mike |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
comment:12
The patch to the Makefile still uses I'd quote |
This comment has been minimized.
This comment has been minimized.
comment:13
Some files from ~/Sage/spkgs/lcalc-1.23.p7$ ls src/include/
cmdline.h Lfind_zeros.h
getopt.h Lgamma.h
Lcommandline_elliptic.h Lglobals.h
Lcommandline_globals.h Lgmpfrxx.h
Lcommandline.h Lgram.h
Lcommandline_misc.h L.h
Lcommandline_numbertheory.h Lint_complex.h
Lcommandline_twist.h Lmisc.h
Lcommandline_values_zeros.h Lnumberzeros.h
Lcommon.h Lnumeric.h
Lcommon_ld.h Lprint.h
Lcomplex.h Lriemannsiegel_blfi.h
Ldirichlet_series.h Lriemannsiegel.h
Ldokchitser.h Lvalue.h
Lexplicit_formula.h Lvalue.h.bak
Lexplicit_formula.h.swap.crap mpfr_mul_d.h I wonder if it's worth to run Lcalc's tests from an I'd really appreciate having the changes to spkgs to be "finally" reviewed committed, especially if files get deleted, as is the case here. Also, IMHO commit messages are subject to review as well, no matter if some merge script would add a generic one in case it is missing. |
Changed reviewer from Volker Braun to Volker Braun, Leif Leonhardy |
comment:14
New spkg, same location. This cases care of most of leif's comments. I believe that the remainder of leif's comments should not prevent a positive review for this ticket. About committing the changes: I find it very annoying to have to commit the changes everytime I make a change. If you are happy with the spkg except for the committing of the changes, I can still commit the changes at that point. |
This comment has been minimized.
This comment has been minimized.
comment:15
Replying to @jdemeyer:
You don't have to, unless you publish them. ;-) And you can always keep an uncommitted copy. I don't think it's much work to commit the changes if you have to create (perhaps upload) and announce a new spkg anyway.
Well, not committing the changes means I have to commit your changes in your name (with some random commit message) and repackage the spkg
The first point is of course more crucial, since not only lazy people might just install and test the spkg as is, such that potential errors will only show up much later. So not committing the changes is IMHO simply bad practice unless the spkg is really and explicitly said to be in some alpha state, most probably not ready and not intended to get merged as is (despite needing review / testing by others), which might be ok in rare cases. There's a reason the term commit is used rather than just e.g. save changes, the former implying some confidence or conviction by the author (or committer). |
comment:16
Replying to @nexttime:
This is simply false. Committing does not delete the files. Committing only changes files inside |
comment:17
Replying to @nexttime:
When doing this, committing the changes is probably the least of your worries. Basing a new spkg on a not-yet-positively-reviewed skpg is not a very good idea anyway (see #11605 vs #11130). |
comment:18
From first glance:
# If SAGE_DEBUG is set either unset (the default), or set to 'yes'
...
echo >&2 "Code will be built with debugging information present. Set 'SAGE_DEBUG' to 'no' if you don't want that."
# Actually anything othe than 'yes' will cause
# no debugging information to be added. (As mentioned, I also would add
The rest looks ok. I've btw. successfully tested the previous p7 with PARI 2.5.0.p0 and Sage 4.7.1.rc0 (Ubuntu 10.04 x86_64, GCC 4.5.1), both with our current MPIR and the newer MPIR 2.1.3 from #8664. |
comment:37
As said before, I don't see much reason to support older versions of PARI. The code would only get more complicated without much gain. |
comment:38
leif, do you still plan to make a follow-up spkg? I don't mind making a new spkg myself based on your suggestions, I just need to know how to move ahead. |
comment:39
Replying to @jdemeyer:
I'll look at it right now, as I don't exactly recall what the remaining changes were. I still don't think we should break Lcalc compatibility with older PARI versions for no reason; the change to make it work with both is quite trivial, and we should submit it upstream anyway. |
comment:40
Replying to @nexttime:
Oh, it seems all my changes are already committed (July 30th btw.), unfortunately all in yet another p8. I'll try to separate them into a p9, then you can make whatever changes you want (e.g. reverting the removal of overlong lines), and provide a p10. IMHO it's very convenient to have at least an overview of the patches in |
comment:41
Replying to @nexttime:
My apologies, I got confused with the pari spkg, which does have a |
comment:42
Replying to @nexttime:
This seems to still apply. I guess this was what was delaying my spkg, i.e., why I didn't upload it.
Not sure about that, but I presumably already did; will check this. |
This comment has been minimized.
This comment has been minimized.
comment:45
New spkg containing most (not all) of leif's changes: Needs review. |
Attachment: lcalc-1.23.p8-p9.diff.gz Attachment: lcalc-1.23.p6-p9.diff.gz |
Changed author from Jeroen Demeyer to Jeroen Demeyer, Leif Leonhardy |
This comment has been minimized.
This comment has been minimized.
comment:48
I read through the diff to the spkg, and everything looked reasonable to me. I applied the new pari stuff (#11330), then build this spkg, and it worked fine, and passed the full test suite (for Sage) after applying it. I tried one simple example "by hand" that actually uses lcalc, and was not pleased by what happened:
Basically, every time you use lcalc to do anything with elliptic curve L-series, you get some mysterious warning (of course, really output from PARI). However, the problem has been in released Sage for a long time, so it should be a separate ticket. It was in sage-4.7. That's now #11985. So I say: positive review for this ticket. Great work guys for cleaning up all kinds of little issues. This is not an easy spkg, to put it mildly. |
Changed reviewer from Volker Braun, Leif Leonhardy to Volker Braun, Leif Leonhardy, William Stein |
comment:49
Thanks a lot for the review. |
Milestone sage-4.7.3 deleted |
Merged: sage-4.8.alpha1 |
PARI version 2.4.4 (see #11130, now aiming at 2.5.0) no longer has an
allocatemoremem()
function, whichlcalc
uses. There is a functionallocatemem()
which can be used instead.The second reviewer patch (not in the spkg) removes the requirement for a PARI >= 2.4.4 spkg, i.e. with that applied it will build with our current ones as well.
New spkg: http://boxen.math.washington.edu/home/jdemeyer/spkg/lcalc-1.23.p9.spkg (this will only build after installing the PARI spkg from #11130).
Reviewers should look at attachment: lcalc-1.23.p6-p9.diff to see the difference with the lcalc spkg currently in Sage.
Depends on #11130
Upstream: Reported upstream. Developers acknowledge bug.
CC: @kiwifb @rishikesha
Component: packages: standard
Keywords: lcalc spkg
Author: Jeroen Demeyer, Leif Leonhardy
Reviewer: Volker Braun, Leif Leonhardy, William Stein
Merged: sage-4.8.alpha1
Issue created by migration from https://trac.sagemath.org/ticket/11321
The text was updated successfully, but these errors were encountered: