[sage-devel] Re: Geo buckets?

[Simon King]

Hi Christian,

On 2012-05-24, ederc  wrote:
> sbuckets are just for bucket sortings.
> kbuckets are singular's implementation of geobuckets, those are used when=
>=20
> it comes to gr=F6bner stuff.

Thank you for the clarification!

Best regards,
Simon

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: Geo buckets?

[ederc]

hi all,

sbuckets are just for bucket sortings.
kbuckets are singular's implementation of geobuckets, those are used when 
it comes to gröbner stuff.

christian

On Thursday, May 24, 2012 5:54:40 AM UTC+2, john_perry_usm wrote:
>
> Simon 
>
> > Is it perhaps the case that kbucket is Singular's notion of geo bucket? 
>
> Now that you mention it, I seem to recall a conversation with one of 
> the Singular developers where he mentioned this. I could try to look 
> through the emails at my office desktop, but it's probably better to 
> ask him directly. I'll do that & tell you the answer. 
>
> john perry

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: Geo buckets?

[john_perry_usm]

Simon

> Is it perhaps the case that kbucket is Singular's notion of geo bucket?

Now that you mention it, I seem to recall a conversation with one of
the Singular developers where he mentioned this. I could try to look
through the emails at my office desktop, but it's probably better to
ask him directly. I'll do that & tell you the answer.

john perry

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: "hash X not found" errors prevent Sage 5.0 from starting on Ubuntu 12.04

[Keshav Kini]

"Peter H."  writes:
> I posted a few days ago at
>  
> http://ask.sagemath.org/question/1432/code-for-hash-md5-was-not-found-after-fresh-sage
> about a problem after Sage 5.0 install on Ubuntu 12.04.  Sage will not
> start at all.
>
> No one has answered it in that forum to suggest any cause or any way
> to investigate the problem.  Does anyone here have any ideas?

>From the traceback (and from the path name) I can see that you are using
the binary that was buit on our build bot "redhawk", which is running
Ubuntu 10.04 LTS. Since your machine is running Ubuntu 12.04, it is not
surprising that the binary you downloaded does not work.

Unfortunately, while it would be nice if we had a larger variety of
operating systems on our build bots, we do not currently happen to have
one running 64-bit Ubuntu 12.04, only one running 32-bit Ubuntu 12.04.
Please try to install from source (following instructions in the
documentation section "Install from Source Code"). This will take a
relatively long time, so consider running it overnight if you can.

Another possible alternative would be to try running the 32-bit Ubuntu
12.04 binary, but I don't know if it would work. Can anyone confirm?

-Keshav


Join us in #sagemath on irc.freenode.net !

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Re: [sage-devel] Re: weighted projective spaces

[David Eklund]

Hi,

I guess Volker meant Riemann surface when saying smooth elliptic surface in 
P^2 (that is, an elliptic curve). But as Marco says, this is about 
hyperelliptic curves. 

Volkers comment that toric varieties in Sage are often assumed to be 
defined over a field is important and has implications for the issues 
regarding hyperelliptic curves discussed above. Developing toric varieties 
over general rings sounds like an intriguing (and elaborate) project. I 
think I will start small by adding explicit construction of weighted 
projective spaces to the toric variety library. We'll see what happens 
after that.

Marco's standard solution number 1) with two glued A^2 also sounds useful. 
Maybe this is similar to what we would actually be doing with the weighted 
projective plane P(1,g+1,1) where g is the genus of the curve.

best
/David

On Tuesday, May 22, 2012 3:39:34 PM UTC-6, Marco Streng wrote:
>
> Op 22-05-2012 15:26, Volker Braun schreef: 
> > On Tuesday, May 22, 2012 4:16:08 AM UTC-4, Marco Streng wrote: 
> > 
> > Definitely! That would make it possible to have a smooth projective 
> > model, with the correct points at infinity. 
> > 
> > 
> > I don't understand that sentence - a smooth elliptic surface can already 
> > be embedded in P^2, right? 
>
> That sentence refers to the sentence by David Eklund right above it. So 
> it has nothing to do with elliptic surfaces, but is about hyperelliptic 
> curves. 
>
> Hyperelliptic curves can be mapped birationally onto a curve in P^2, 
> just by going via the standard model in A^2 of the form y^2 + h(x)*y = 
> f(x). But the image is not smooth at infinity for any hyperelliptic 
> curve (of genus >=2). To make the image smooth, the standard solutions 
> are to 
> 1) glue two copies of A^2, 
> 2) use a higherdimensional P^n, or 
> 3) use a weighted projective 2-dimensional space. 
>
> I think David was aiming at (3), and I was simply welcoming that. 
>
> > 
> > Note that the toric variety code assumes that the base ring is a field 
> > at various places. So for number theory purposes it might be good to 
> > split things into ToricVariety_ring vs. ToricVariety_field. Its mostly 
> > my ignorance about toric varieties over general rings that prevented me 
> > from doing so... 
>
> > 
> > 
> > -- 
> > To post to this group, send an email to sage-devel@googlegroups.com 
> > To unsubscribe from this group, send an email to 
> > sage-devel+unsubscr...@googlegroups.com 
> > For more options, visit this group at 
> > http://groups.google.com/group/sage-devel 
> > URL: http://www.sagemath.org 
>
>

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] "hash X not found" errors prevent Sage 5.0 from starting on Ubuntu 12.04

[Peter H.]

I posted a few days ago at
 
http://ask.sagemath.org/question/1432/code-for-hash-md5-was-not-found-after-fresh-sage
about a problem after Sage 5.0 install on Ubuntu 12.04.  Sage will not
start at all.

No one has answered it in that forum to suggest any cause or any way
to investigate the problem.  Does anyone here have any ideas?

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: Sage 5.0 on ARM

[mmarco]

I did build the chroot image, but when i test it sage fails at startup
because gap cannot load several packages. It seems that they are not
installed. I suspect that those packages are missing. Does the tarball
that you uploaded work for you on its own?

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: Geo buckets?

[Simon King]

Dear John, dear Martin,

On 2012-05-23, Martin Albrecht  wrote:
> On Wednesday 23 May 2012, john_perry_usm wrote:
>> You probably knew this already, but: peeking at Singular's source, I
>> find at least two different bucket structures in Singular: something
>> called, kbuckets, and something called sbuckets. The include file for
>> 
>> the latter has this information:
>> > Purpose: declaration of routines for sorting and adding up polys using
>> > 
>> >   a bucket sort
>> > 
>> > Note:If you need to extract the leading momonial of a bucket,
>> > 
>> >   use kbuckets, instead.
>
> Doesn't that just refer to bucket sorting, i.e., not to geo buckets at all?

That was my first thought as well. But after all, "extract leading
monomial of a bucket" pretty much sounds like the purpose of geo buckets
(namely, speed up the summation of a long list of relatively small polynomials,
where the items on the list are not known in advance but are determined
by the leading monomials of partial sums).

Is it perhaps the case that kbucket is Singular's notion of geo bucket?

> The closest we come to using them directly is in the groebner_strategy.pyx 
> file.

OK, I'll have a look at that. Thank y'all!

Simon


-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Re: [sage-devel] Re: Geo buckets?

[Martin Albrecht]

On Wednesday 23 May 2012, john_perry_usm wrote:
> Simon
> 
> On May 22, 9:15 am, Simon King  wrote:
> > AFAIK, Singular uses the geo bucket data structure for internally
> > representing polynomials during Gröbner basis computations.
> 
> You probably knew this already, but: peeking at Singular's source, I
> find at least two different bucket structures in Singular: something
> called, kbuckets, and something called sbuckets. The include file for
> 
> the latter has this information:
> > Purpose: declaration of routines for sorting and adding up polys using
> > 
> >   a bucket sort
> > 
> > Note:If you need to extract the leading momonial of a bucket,
> > 
> >   use kbuckets, instead.

Doesn't that just refer to bucket sorting, i.e., not to geo buckets at all?
 
> I don't know the difference between the two, otherwise. Neither
> appears to be wrapped directly in Sage, judging from
> mpolynomial_libsingular.pxd.

The closest we come to using them directly is in the groebner_strategy.pyx 
file.

Cheers,
Martin

--
name: Martin Albrecht
_pgp: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x8EF0DC99
_otr: 47F43D1A 5D68C36F 468BAEBA 640E8856 D7951CCF
_www: http://martinralbrecht.wordpress.com/
_jab: martinralbre...@jabber.ccc.de

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Re: [sage-devel] ZODB in sage

[Julien Puydt]

Le lundi 21 mai, Julien Puydt a écrit:
> Isn't it ugly to work around that setuptools problem by tainting the
> dependencies? Perhaps the following scheme would work too, with the
> advantage that it would clearly separate on the one hand the correct
> dependencies between the packages and on the other hand the workaround
> to serialize for setuptools installations :
> 
> setuptools_serializer_stamp_1: $(INST)/$(FIRST)
>   touch $@
> $(INST)/$(SECOND): setuptools_serializer_stamp_1
> 
> setuptools_serializer_stamp_2: $(INST)/$(SECOND)
>   touch $@
> $(INST)/$(THIRD): setuptools_serializer_stamp_2

This is now a ticket with an attached patch :

http://trac.sagemath.org/sage_trac/ticket/12994

Snark on #sagemath

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Re: [sage-devel] Re: C build problems on MacOS 10.7 Lion.

[Dima Pasechnik]

On Wednesday, 23 May 2012 16:42:24 UTC+2, John H Palmieri wrote:
>
>
>
> On Tuesday, May 22, 2012 11:50:26 PM UTC-7, Jeroen Demeyer wrote:
>>
>> On 2012-05-23 03:14, John H Palmieri wrote: 
>> > Try installing the command line tools and see if that helps. Sage 
>> > bundles GCC, but it doesn't necessarily include everything which is 
>> > installed with the command line tools, so the machine might be missing 
>> > some important component. 
>> What are these "command line tools"?  The problem is a missing include 
>> file. 
>>
>
> They are an optional part of Xcode: they allow you to run gcc and other 
> programs from the command-line, rather than just through Apple's Xcode app. 
> So installing them puts various files into standard places (e.g., gcc goes 
> into /usr/bin). I don't know if it also puts libraries or headers in 
> standard places.
>
> As far as I know, you do not need Xcode app any more! See 
http://kennethreitz.com/xcode-gcc-and-homebrew.html
So you can download these command line tools for free 
from http://developer.apple.com/downloads
(with a free registration)
With this, why would one need Xcode app for Sage any more?
(well, this is getting slightly off-topic ––  so why won't we require these 
tools for installing Sage from Source on OSX 10.7 ?)

Dima

 

> -- 
> John
>
>

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: sagenb.org

[Jason Grout]

On 5/23/12 6:26 AM, John Cremona wrote:

I wonder how many of those 9000 accounts are still active.  COuld some
of them be deleted, or at least mothballed?




There's one more 0 there...91,000 accounts.




I was thinking aboutthis for our own server, where we only have 75
users, at least half of which are past students who have left and
could not log in even if they wanted too.  I would like to be able to
delete their accounts.  What actually happens when I (as admin users)
click on "suspend" by the username in the user admin page?  And is
there any easy way to see which users have not logged in for (say) the
last year, or 6 months?  I guess that

  ls -lt /home/sage/.sage/sage_notebook.sagenb/home/

would be one way?



Yes, that would be one way.  We don't have a good system for migrating 
users off of a server.  It would be good if there was an easy way to 
backup those accounts, delete them, and then easily restore them if needed.


On sagenb.org, if I do

cd home/
ls -lasHt .

then the 2011 accounts start at around line 16,000, and the 2010 
accounts start at around 76,000.  So if kept accounts going back to 2010 
or so, that would still be 76,000 accounts.  Most of our accounts have 
been created recently.


Thanks,

Jason

--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Re: [sage-devel] Re: C build problems on MacOS 10.7 Lion.

[John H Palmieri]

On Tuesday, May 22, 2012 11:50:26 PM UTC-7, Jeroen Demeyer wrote:
>
> On 2012-05-23 03:14, John H Palmieri wrote: 
> > Try installing the command line tools and see if that helps. Sage 
> > bundles GCC, but it doesn't necessarily include everything which is 
> > installed with the command line tools, so the machine might be missing 
> > some important component. 
> What are these "command line tools"?  The problem is a missing include 
> file. 
>

They are an optional part of Xcode: they allow you to run gcc and other 
programs from the command-line, rather than just through Apple's Xcode app. 
So installing them puts various files into standard places (e.g., gcc goes 
into /usr/bin). I don't know if it also puts libraries or headers in 
standard places.

-- 
John

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: Geo buckets?

[john_perry_usm]

Simon

On May 22, 9:15 am, Simon King  wrote:
> AFAIK, Singular uses the geo bucket data structure for internally
> representing polynomials during Gröbner basis computations.

You probably knew this already, but: peeking at Singular's source, I
find at least two different bucket structures in Singular: something
called, kbuckets, and something called sbuckets. The include file for
the latter has this information:

> Purpose: declaration of routines for sorting and adding up polys using
>   a bucket sort
> Note:If you need to extract the leading momonial of a bucket,
>   use kbuckets, instead.

I don't know the difference between the two, otherwise. Neither
appears to be wrapped directly in Sage, judging from
mpolynomial_libsingular.pxd.

john perry

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: sage mirror path?

[Harald Schilly]

Hello, the path to the rsync master is

boxen.math.washington.edu::sage

See http://wiki.sagemath.org/MirrorNetwork

Harald




On Tue, May 22, 2012 at 11:45 PM, Dima Pasechnik  wrote:

>
>
> On Tuesday, 22 May 2012 21:00:57 UTC+2, Nils Bruin wrote:
>>
>> I've discovered that the sage mirror www.cecm.sfu.ca/sage isn't
>> updating anymore. It's doing a cron job trying to rsync with
>>
>> cecmir...@boxen.math.**washington.edu:/home/was/www/**sage/
>>
>> which fails. Has there been a change to the path required to sync to?
>> (It hasn't able to find this directory since Fri Apr 27 04:30:01 PDT
>> 2012)
>>
>> same applies to the Sage mirror at NTU (Singapore)
> By the way, I have a complaint: the mirror info says:
>
> Nanyang Technological University
> Last Updated by 
> HaraldS on
> May 15, 2011
>  http://jambu.spms.ntu.edu.sg/sage/index.html
> Nanyang Business 
> School
>36 
> reviews
> Nanyang Ave
> Singapore 639798
> 9847 5423
> ntu-singapore.com/nbs
>
> But it is not by NBS, it's by SPMS, School of Physical and Mathematical
> Sciences:
> http://www.spms.ntu.edu.sg/
>
>
>
>> Cheers,
>>
>> Nils
>
>

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Re: [sage-devel] maxima error on calls like solve((A+p^B,p), is that the intended behavior?

[Jonathan]

It is Jonathan G and I will try to open a ticket. I'm running 3 workshops 
on computer software and hardware for our Chemistry, Physics and Math 
departments over the next few days, so we'll see when I can squeeze it in.

I agree that maybe forcing Maxima to do purely symbolics might be the 
cleanest solution.

Jonathan

On Tuesday, May 22, 2012 11:55:20 AM UTC-5, Benjamin Jones wrote:
>
> On Tue, May 22, 2012 at 8:27 AM, Jonathan  wrote: 
> > Yes, I made a typo..Thanks for realizing what I meant. 
> > 
> > However, I still think that a solution that has complex answers is OK. 
>  If 
> >  you are finding the symbolic solution isn't it better to find the 
> general 
> > solution and then deal with specific cases?  I don't see what is wrong 
> with: 
> > p = (-A)^(1/B), even when B = 2 and A > 0.  This just means p = 
> > (+/-)sqrt(-A). I just don't see any problem with having two possible 
> > answers, even if they are complex.  Is the problem that Maxima is trying 
> to 
> > solve the problem restricting it to RR? 
> > 
> > I believe this is a case where Sage should try to catch this kind of 
> Maxima 
> > error and allow the user to get the general answer without having to 
> drop 
> > into Maxima. 
> > 
> > So, I am arguing that Sage really should return: p = (-A)^(1/B) in cases 
> > like this and that when Maxima asks a question, we need a better 
> explanation 
> > for the user of how to respond. 
> > 
> > Jonathan 
> > 
>
> Well, I agree. I guess I was trying to say that a solution, if one 
> exists, depends on the context (the underlying ring and also the 
> values of A and B) so I don't think it's unreasonable for Maxima to 
> ask for more information before giving an apparent solution which may 
> or may not really exist. 
>
> Mathematica seems to agree with you to: 
>
> http://www.wolframalpha.com/input/?i=solve+A+%2B+p%5EB+%3D+0+for+p 
>
> I definitely agree that Sage could be more helpful here as the middle 
> man. I wonder if Maxima has any flags related to this, e.g. 
> `return_purely_symbolic_solutions = true` would return `[ p = 
> (-A)^(1/B) ]` and we could wrap this flag as an option to solve? 
>
> -- 
> Benjamin Jones 
>

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Re: [sage-devel] Re: sagenb.org

[John Cremona]

I wonder how many of those 9000 accounts are still active.  COuld some
of them be deleted, or at least mothballed?

I was thinking aboutthis for our own server, where we only have 75
users, at least half of which are past students who have left and
could not log in even if they wanted too.  I would like to be able to
delete their accounts.  What actually happens when I (as admin users)
click on "suspend" by the username in the user admin page?  And is
there any easy way to see which users have not logged in for (say) the
last year, or 6 months?  I guess that

 ls -lt /home/sage/.sage/sage_notebook.sagenb/home/

would be one way?

John

On 23 May 2012 11:56, Jason Grout  wrote:
> On 5/23/12 5:45 AM, P Purkayastha wrote:
>>
>> Thanks. It is snappier for sure. I think something is wrong with the
>> openid authentications. When I try to login via google/yahoo, I get an
>> error just by clicking on the corresponding icon:
>>
>> ||
>> ProxyError
>>
>> Theproxy server received an invalid response froman upstream server.
>> Theproxy server could nothandle the request GET /loginoid.
>>
>> Reason:Errorreading fromremote server
>>
>>
>> Normal username logins are working.
>
>
>
> I see that too.  I've restarted the servers to use twisted for the time
> being until someone figures out a solution for uwsgi.
>
>
> Thanks,
>
> Jason
>
> --
> To post to this group, send an email to sage-devel@googlegroups.com
> To unsubscribe from this group, send an email to
> sage-devel+unsubscr...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/sage-devel
> URL: http://www.sagemath.org

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: sagenb.org

[Jason Grout]

On 5/23/12 5:45 AM, P Purkayastha wrote:

Thanks. It is snappier for sure. I think something is wrong with the
openid authentications. When I try to login via google/yahoo, I get an
error just by clicking on the corresponding icon:

||
ProxyError

Theproxy server received an invalid response froman upstream server.
Theproxy server could nothandle the request GET /loginoid.

Reason:Errorreading fromremote server

Normal username logins are working.



I see that too.  I've restarted the servers to use twisted for the time 
being until someone figures out a solution for uwsgi.


Thanks,

Jason

--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: sagenb.org

[P Purkayastha]

Thanks. It is snappier for sure. I think something is wrong with the openid 
authentications. When I try to login via google/yahoo, I get an error just 
by clicking on the corresponding icon:

Proxy Error

The proxy server received an invalid response from an upstream server.
 The proxy server could not handle the request GET /loginoid.

 Reason: Error reading from remote server

Normal username logins are working.

On Wednesday, May 23, 2012 6:19:37 PM UTC+8, jason wrote:
>
> On 5/18/12 12:08 AM, Jason Grout wrote: 
> > I upgraded the Sage servers to 5.0, and it seems that there is some sort 
> > of really bad scalability problem that I'm only seeing on sagenb.org(or 
> > more likely, is only noticeable on a server the size of sagenb.org). 
>
> I found the problem: 
>
>
> https://github.com/sagemath/sagenb/commit/8ff47fdc4e06cdf75c9a56a5528310833b3acdf4
>  
>
> Basically, every single time the home page of the server was shown, the 
> server was going through the entire list of usernames on the server and 
> doing a string compare of each of those names with 4 strings.  Wow. 
> Each homepage render was taking 1/3 second, so it only took a few 
> simultaneous requests to peg the server at 100% CPU.  This apparently 
> has been creeping up on us, and now that sagenb.org has over 91,000 
> accounts, it was swamping everything else. 
>
> Anyways, it's fixed.  And now sagenb.org and *.sagenb.org are running 
> uwsgi to boot.  Things feel pretty snappy now.  Please let me know if 
> you see any problems. 
>
> Thanks, 
>
> Jason 
>
>

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Re: sagenb.org

[Jason Grout]

On 5/18/12 12:08 AM, Jason Grout wrote:

I upgraded the Sage servers to 5.0, and it seems that there is some sort
of really bad scalability problem that I'm only seeing on sagenb.org (or
more likely, is only noticeable on a server the size of sagenb.org).


I found the problem:

https://github.com/sagemath/sagenb/commit/8ff47fdc4e06cdf75c9a56a5528310833b3acdf4

Basically, every single time the home page of the server was shown, the 
server was going through the entire list of usernames on the server and 
doing a string compare of each of those names with 4 strings.  Wow. 
Each homepage render was taking 1/3 second, so it only took a few 
simultaneous requests to peg the server at 100% CPU.  This apparently 
has been creeping up on us, and now that sagenb.org has over 91,000 
accounts, it was swamping everything else.


Anyways, it's fixed.  And now sagenb.org and *.sagenb.org are running 
uwsgi to boot.  Things feel pretty snappy now.  Please let me know if 
you see any problems.


Thanks,

Jason

--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

[sage-devel] Support sage --sdist sage-VERSION (#6338) needs review

[Jeroen Demeyer]

For people who forget how to call sage --sdist or --bdist, we should allow

sage --sdist sage-VERSION

in addition to

sage --sdist VERSION

By stripping of "sage-" from the version number if given.

There is a patch for this at
http://trac.sagemath.org/sage_trac/ticket/6338


Please review,
Jeroen.

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org