Scientific Repository?

IncredibleLaser · 2010-11-22 12:34:12

Hi,
I recently have been thinking about a scientific repository for Arch Linux that could host all scientific packages in the AUR including their dependencies. Stefan Husmann maintains a large number of packages in the AUR that I use and he also maintains many more packages people might be interested in. aurperl (a bot as far as I understand) provides many dependencies. If anyone is interested, especially the people who already maintain these packages, I'd love to see a repository dedicated to these packages. I know a few people at my university who use ubuntu and always complain about the old version of texlive. Arch is already leading in this regard and without the need to compile very specific packages, it might become more attractive for everyone seeking an up-to-date distribution. If enough volunteers are found, I'd be willing to fund the server needed (I've been thinking about this one) for at least one year. I'd also like to hear your comments about the needed server capacity.
Best regards

ngoonee · 2010-11-22 12:55:37

The thing about science packages is that they're usually REALLY LARGE, so bandwidth requirements are high. This is just my uninformed opinion, and is almost solely based on the crazy size required by sage...

Allan · 2010-11-22 13:03:15

They are not all large...

allan@mugen /var/abs/local/genetics 
> du -h
584K    ./qtdt
3.9M    ./pap
9.1M    ./haploview
384K    ./pedstats
260K    ./pedcheck
168K    ./crimap
6.5M    ./mendel-bin
720K    ./pedfiddler
700K    ./ms
2.5M    ./merlin
9.4M    ./plink
34M    .

And that is source and package files. I think the issues is more that they are low usage... I think not a single one of those packages are in the AUR indicating their user base in near one.

karol · 2010-11-22 14:18:21

I think sage is a really special package - in many regards. Maybe you should talk to its maintainer, Thomas Dziedzic. There was much talk about it in the ML.
http://www.archlinux.org/packages/commu … thematics/

ConnorBehan · 2010-12-10 10:28:51

Allan wrote:

I think not a single one of those packages are in the AUR indicating their user base in near one.

You could test that assumption by putting them in the AUR and seeing if the "votes" increase!

jelly · 2010-12-13 13:05:41

Hi, you could also start setting up an FTP server and getting some math packages in there See how popular it is and work further on it.

kaitocracy · 2010-12-14 00:12:52

Does anyone have a disk space or bandwidth estimate? If it's reasonable I'll host it. I've got the build system pretty much set up already.

Last edited by kaitocracy (2010-12-14 00:18:06)

Stefan Husmann · 2010-12-14 15:51:55

Before estimating the size: Is there a list of packages that are candidates?

IncredibleLaser · 2010-12-14 19:14:32

Sorry for not posting, I have an exam tomorrow.

tomd123 · 2010-12-15 01:11:50

Hello, I also would like to know what the list of candidates would be? just curious

BTW, my original interest was in science/engineering/math apps (and still is) but since starting as a TU I have taken over maintenance of Ghost1227's packages because he hasn't maintained them lately.

IncredibleLaser · 2010-12-15 09:26:49

Candidates would be LaTeX packages (I for one miss biblatex), but anything scientific might be added. As an electrical engineer, I sometimes use kicad which is "missing" from the official repositories. There might be many more packages that are useful for other fields that I don't know of.

kaitocracy · 2010-12-15 09:43:00

I'm going to need an actual list if I'm going to be able to do anything about this.

IncredibleLaser · 2010-12-15 13:46:55

I use the following packages:
- biblatex
- gnuplot-cvs for tikZ output (will be included in 4.5)
- kicad
- octave-forge
- pdftk

These are the packages I use. I don't know if other people would appreciate such a repository, that's why I started the thread. There are some packages in the AUR in the scientific category that have quite a lot of votes.

linux-ka · 2010-12-15 15:01:34

well...beside octave and gnuplot-cvs, maybe the following:
- qtiplot
- gwyddion
- gnuplot-py
- gnuplot-cvs-pdflib (printing directly to pdf is more comfortable and evoids conflicts due to the often missing epstopdf)
- allegro
- cython
- imagemagick
- gimp
- maxima
- python-numeric and every other extension used for plotting/data-analysis (scipy,numpy, etc)

These are often used by my colleagues and myself .

IncredibleLaser · 2010-12-15 16:39:19

Many of your packages are already in [extra] or [community].

linux-ka · 2010-12-15 17:57:21

of cause, but actually every mentioned package before is available if I guess right .

So, wasn't it to point out what could be included into a scientific repo?

IncredibleLaser · 2010-12-15 22:33:23

Well, it doesn't make sense to include a package in the scientific repositories if it is available in one of the standard repositories.

ConnorBehan · 2010-12-16 20:53:49

Some scientific packages not in official repos:

-calculix
-elmer
-grace-openmotif (in AUR)
-chemtool-proper (in AUR)
-xephem
-pymol
-jmol / java-cdk
-rasmol (used to be there but horribly buggy)
-BLAST
-folding@home / rosetta@home / mprime
-openaxiom
-yacas
-euler
-brlcad
-freecad
-openscad
-GEDA

And I don't even think I mentioned any obscure ones. Plus there's the list of packages Allan mentioned. One thing you could do is put extra Celestia content in the repos. I started off doing this in the AUR, but with the way KMS is going, I can't really use Celestia without it crashing so there's not much point in my updating the packages. The Arch repos include some things from CTAN and CPAN so why not CRAN? Here are the R packages I installed using one of the automated CRAN scripts floating around the forums:

[root@connors-laptop ~]# pacman -Q | grep cran
r-cran-abind 1.1.0-1
r-cran-aplpack 1.2.3-1
r-cran-car 1.2.16-1
r-cran-colorspace 1.0.1-1
r-cran-effects 2.0.10-1
r-cran-hmisc 3.7.0-1
r-cran-leaps 2.9-1
r-cran-lmtest 0.9.26-1
r-cran-multcomp 1.1.6-1
r-cran-mvtnorm 0.9.9-1
r-cran-rcmdr 1.5.4-1
r-cran-relimp 1.0.1-1
r-cran-rgl 0.91-1
r-cran-zoo 1.6.2-1

Hope that helps!

Julius2 · 2010-12-17 13:57:50

I know a few people at my university who use ubuntu and always complain about the old version of texlive.

I wouldn't necessarily call TeX a "scientific application", although I have yet to see anyone in the arts or humanities use it. Are the scientific applications going to be software commonly used by people in the sciences, or things that are explicitly scientific applications?

kaitocracy · 2010-12-18 23:45:00

I can go ahead and host this. Send me a PM if you're interested in working on this. It'll be hosted on a Linode 1536 in Newark, New Jersey. A full development stack is already set up.

ConnorBehan · 2010-12-19 05:26:45

Julius2 wrote:

I wouldn't necessarily call TeX a "scientific application", although I have yet to see anyone in the arts or humanities use it.

Those people are rare but they exist.

Pank · 2010-12-19 11:36:58

ConnorBehan wrote:

Julius2 wrote:
I wouldn't necessarily call TeX a "scientific application", although I have yet to see anyone in the arts or humanities use it.
Those people are rare but they exist.

I don't know if LaTeX usage is even rare. The applications of TeX within the arts are, I would argue, substantial. Check the TUG showcase or the very impressive samples here.

Anyhow, the only reason for including e.g. TeXlive in an Arch-specific science repository should be to offer something different from the offerings of extra. For example a minimal package relying solely on tlmgr would be nice.

There is also a rather new ConTeXt package, which might be valuable in such a repository.

mike_r · 2010-12-20 01:16:25

Why not setup up a really good wiki page describing the resources available for scientific computing? I think the bigger "problem" is that people do not realize what is actually available. I had intended to write a "Scientific Computing with Arch" article for the monthly newsletter before reality trampled the garden of good intentions, but I personally find the python/cython/numpy/scipy/matplotlib etc. stack extremely powerful, and it is already all in the community repository. Add a few things like pyfits from AUR, and you've got the software I use professionally every day.

So rather than Yet-Another-Repository, why not assemble a decent User Guide on the wiki?

Mike

Mr.Elendig · 2010-12-20 16:05:08

Here is my take on this:

The best way to get a [scientific] repo up is to stop talking about it and just set it up (as a 3.rd party repo) instead. That is the most efficient way to get something done. If it turns out to be a sucsess then it might even turn out to become a official repo. After all, that is how the x86_64 fork of arch started

Last edited by Mr.Elendig (2010-12-20 16:06:13)

jelly · 2010-12-21 11:44:35

Mr.Elendig wrote:

Here is my take on this:
The best way to get a [scientific] repo up is to stop talking about it and just set it up (as a 3.rd party repo) instead. That is the most efficient way to get something done. If it turns out to be a sucsess then it might even turn out to become a official repo. After all, that is how the x86_64 fork of arch started

I completely agree with MrElendig, making an arch repo is easy and if it's a succes maybe it will become official.

If you need help setting up a repo, look at Arch-Games: https://github.com/Arch-Games/arch-games we have some nice scripts to check packages with there AUR versions etc.

Arch Linux

#1 2010-11-22 12:34:12

Scientific Repository?

#2 2010-11-22 12:55:37

Re: Scientific Repository?

#3 2010-11-22 13:03:15

Re: Scientific Repository?

#4 2010-11-22 14:18:21

Re: Scientific Repository?

#5 2010-12-10 10:28:51

Re: Scientific Repository?

#6 2010-12-13 13:05:41

Re: Scientific Repository?

#7 2010-12-14 00:12:52

Re: Scientific Repository?

#8 2010-12-14 15:51:55

Re: Scientific Repository?

#9 2010-12-14 19:14:32

Re: Scientific Repository?

#10 2010-12-15 01:11:50

Re: Scientific Repository?

#11 2010-12-15 09:26:49

Re: Scientific Repository?

#12 2010-12-15 09:43:00

Re: Scientific Repository?

#13 2010-12-15 13:46:55

Re: Scientific Repository?

#14 2010-12-15 15:01:34

Re: Scientific Repository?

#15 2010-12-15 16:39:19

Re: Scientific Repository?

#16 2010-12-15 17:57:21

Re: Scientific Repository?

#17 2010-12-15 22:33:23

Re: Scientific Repository?

#18 2010-12-16 20:53:49

Re: Scientific Repository?

#19 2010-12-17 13:57:50

Re: Scientific Repository?

#20 2010-12-18 23:45:00

Re: Scientific Repository?

#21 2010-12-19 05:26:45

Re: Scientific Repository?

#22 2010-12-19 11:36:58

Re: Scientific Repository?

#23 2010-12-20 01:16:25

Re: Scientific Repository?

#24 2010-12-20 16:05:08

Re: Scientific Repository?

#25 2010-12-21 11:44:35

Re: Scientific Repository?

Board footer