Previously on AstroBetter, we’ve discussed what operating systems are used in our profession, in particular relative numbers of OS X (Mac) versus Linux users. While it’s good for us at AstroBetter to know our readership, we can use Google Analytics for that. It’s more important for the astronomical community to know the broader landscape, so that as astronomers develop software tools, they are aware of the platforms colleagues will use to access those tools.
So I asked the folks at STScI who run the Astronomer’s Proposal Tool (APT). They’ve been keeping track of what operating system was used to submit every Hubble proposal for the last 7 proposal cycles. They kindly sent me a chart to share. Here it is,
regraphicked for clarity. The Y axis is the percentage* of proposals per year submitted with a given operating system.
Each cycle had between 700 and 1100 proposals submitted. While there may be wavelength-dependent trends, I would argue that Hubble users are a broad cross-section into the astronomical community.
So this is a fascinating chart! Linux has slowly lost market share, and now serves a quarter of users. And check out the decline of Sun, and the corresponding rise of Macs. These are trends we all know — but it’s neat to see quantification.
Comments? Discussion?
* Ignore the small not-summing-to-100% problem; I digitized the charts from powerpoint figures, and didn’t click with fantastic precision.
I really don’t get the Mac fetish among astronomers. It seems kind of tasteless in a way. Generally speaking most people are not flush with cash and we rely on the goodwill of taxpayers, hopefully convincing them that while what we do has few visible economic benefits for them, we’re all interested in the mysteries of the universe and it’s just cool to be able to understand in some sense why we are here. And then people are blowing that generously given taxpayer’s money on ostentatious toys like Macbook Airs, which cost twice as much as a PC laptop and if they have any advantage, it’s not commensurate with the price difference.
I use linux and I have never noticed any reason why a Mac should be better, for work purposes. For what reason should astronomers expect taxpayers to fund their Mac habit?
I disagree quite strongly. It is true that Apple computers cost more than a standard Linux machine. I know when I bought my latest Apple workstation, it was about 30-40% more than expensive than a comparable spec-ed Linux computer. So, naively, you might think that was a waste of money. Well, due the far more complete and stable nature of OS X, I spend far less time doing basic sys-admin stuff — that would occupy a significant time under linux. Perhaps Linux has become more stable and user-friendly over the last 2 or so years, but it was those difficulties (that absolutely existed about 5 years ago) which lead to the great migration to OSX. In fact, I wonder if it wasn’t that need for more user-friendly, stable, complete OS which lead Linux to be more “Mac-like” in it’s current state.
In the US it costs the taxpayers approx $100,000/year or more to fund one postdoc’s salary (incl benefits and overhead). Not including travel and other research expenses.
It costs approx $2000 (including overhead) to buy a decent but not top end laptop which will typically last about 3 years. If you cut the cost of the laptop by $500 you save about $750 total over 3 years.
People should use whatever OS makes them happy and more productive. Saving grant money on your computer is in the noise. Operating system advocacy wars are not interesting in my opinion. I do think (as someone who uses both) that there are warning signs for both: (1) the Linux decline indicates both the fact that the Mac price premium is less than it used to be, and that Linux usability/installation issues continue to be a problem; (2) as mentioned below, Apple’s future direction is more towards closed iOS type systems that would not be useful for, say, programming and data reduction. I’m hoping that their graphics/movie customer base keeps them making “real computers.”
Ben, where you say “People should use whatever OS makes them happy and more productive”, I can’t see any reason to have the bit about happiness there. People should use whatever OS makes them more productive. That’s what we’re paid for, right? Even if the extra cost of a mac is not much in comparison to the total cost of a postdoc it’s still an ostentatious and unnecessary extra cost. It appears that way to me, so I’m sure it must appear that way to the taxpayer.
RW: Happiness in what you do and how you do it is one of the most reliable indicators of your productivity. So not only the costs differences are small (and mostly irrelevant as shown above), but you seems to take a very narrow view of research to the point of suggesting we should dictate what tools to use. I strongly disagree with this approach.
Alberto, if your research computer also makes you happy on a personal level, that’s a bonus, but I think you have an obligation to give the taxpayer good value for money. Macs really don’t give good value for money as far as I can see. The claimed advantage is that you spend less time on system maintenance and software installation; I spend almost no time on these. So what is the extra cost of a mac gaining?
I’m ready to be convinced. I just have never seen any tangible reason why a mac should be a better computer for astronomical research than a linux system, and I’ve seen, in fact, the opposite of what you say – people struggling to install the software they want to use on a machine that’s basically not particularly appropriate for scientific research.
Astronomers, of all professions, getting on high horses about saving a couple hundred dollars… amazing. Is this public outrage about ostentatious OS choices anything more than projection, or is there some data? I wonder how people feel about paying for IDL, given the cheaper options.
I dont use IDL partly because of the price. Python works well and is free. Moreover my students can also use it without worrying about getting them a license too.
Because Mac is a reliable Unix environment – Simple
I think this is biased because most astronomers submit proposals using their laptop. This may not be the machine on which they do heavy data analysis computation. We still see more Linux than Mac in our software downloads.
Keith, if there a simple way to do this, share a similar graph. I’d hate to see Jane’s graph live a life outside of this blog 🙂
RW: I would have to disagree with most of your comments. I use both Macs and Windows, and I given up on Linux laptops almost 8 years ago, even if I still virtualize Linux a lot. Astronomers don’t have a “habit”, we just like to use the best tool available allows us to run command-line and GUI interfaces on the same system. It’s like for cars, we prefer it to horses, that’s all. Nothing against horses. Seriously, Macs allow me to do my work faster, better and more efficiently since I am not spending days installing packages or dealing with drivers.
Ultimately people should be able to use the most appropriate tool for their research. I think the statistics above simply shows how Linux had failed in some areas.
“It’s like for cars, we prefer it to horses, that’s all. Nothing against horses.”
Please, enough of the loaded statements. They are worse than useless without data to support them.
If you’re suggesting that linux is a horse and mac is a car, I think that’s quite ridiculous. If you want to pursue this kind of analogy it’s more like choosing between a car that gets you to where you need to go, with a car that gets you where you need to go but is more expensive and has a more fashionable brand name.
Ah, there’s statistics, statistics, and lies, right?
Using a browser for a proposal might not tell you enough how people do their work. I still find too many thing unacceptable (filename case for example, lack of ease to install random open source software, confusing darwinports vs. fink vs. apple’s compilers) to even bother with a mac, let alone the price. But maybe i’m more patient with the intricasies of installing linux, but there’s no question about ease of installing the software once linux knows it’s hardware, linux still beats the crap out of all of them. I’d be more worried about the move to tablets (like the iPad) where were moving away from open source or at least in control of your software. We’ve been stooped into a mode that I see has no happy ending for science computing, and we’re moving back to elite (and hence more expensive) computers to do our work.
I have to say that I use both Mac and Linux. For most things I prefer linux. For some things, like web browsing, the Mac is better (but honestly, all computational work is easier on Linux). This thing would count me as a Mac user. I would not. I do agree with the financial aspect raised by RW. My current institution almost forced me to take a Mac, so I did. I will not do it again, but I guess other institutes may treat me similarly.
Same happened to me. They, my scientific advisor and others at my department, basically forced me to move to Mac. When I started resisting they said that I would spend too much time administering the system, because Linux is for hackers and they can’t find PC vendors that sell computers with Linux on them! Later, when I proved to them that Dell and HP sell computers with Linux preinstalled they started telling me that their IT department administers only Macs and Windows. After I proved that I am very proficient at Linux and I can do everything by myself very quickly my scientific advisor came to me and told me that we’re a small community and I can’t be allowed to be the black sheep, so I have to do what others do. At that point I was ready to quit, and I said that I don’t need a Mac or a PC and I will use my personal laptop for my job. Surprisingly, he agreed.
So, you see kids. This has nothing to do with money or administering or anything like that. It’s that we’re all human and we like to make other people obey us and do what we tell them to do.
People “spending days installing packages” is exactly what I see the people doing around me that run Macs! A lot of them even give up at some point and choose to run their work related software in a virtual machine running Linux. It was a long time ago that I had to spend time to install drivers on my Linux laptops, modern distributions like Ubuntu came a long way in the last years!
Don’t get me wrong, I like OS X and for any kind of multimedia related work I would choose it over Linux any day.
Macs have had software called “Boot Camp” that allows users to partition HD and install Windows as a separate OS for a while now (maybe since 2006?). So i can run PC programs on my Mac w/ Windows software. Could this be a factor in the market share gain?
Robert: true. My bias I suppose derives from the fact that I’ve been using a laptop for many years and I have not found a similar system as nice as a mac in the Linux world. For desktops and servers we could argue about it. However, I do like OS X.
Alberto and Russell: I spend very, very little time maintaining my Ubuntu system. I’ve been using Ubuntu for about five years. Updates install themselves, fairly frequently, and I click “OK” to those. Other than that, it runs perfectly. I cannot remember ever having had any difficulty installing the software that I wanted to use. So the claimed advantage of a mac, that its OS is more stable and complete, doesn’t seem like a very significant one. And I do see mac users around me struggling quite frequently to install things. Are there other advantages? Once the software is installed, does it run better on a mac? Are there astronomical packages that can only run on macs?
RW: actually for software I am not aware on anything mayor that does not run on both Linux + OS X. My bad experience (a while back) was with a Linux laptop. Perhaps I should reconsider. I do like the very polished OS X environment and I consider it “almost Linux” in that I can run my scripts/code when needed easily. Can’t say the same for Windows unless Cygwin is installed.
~10% of the astronomers use Windows? Seriously?
Yep. When my kids drop the $300 netbook, I don’t cry; it’s *way* better than a mac at just working when plugged into a projector; and it runs chrome and vncviewer just fine (though I’ll admit the putty step is an annoyance).
Wow, this is an old thread, and I know I am in the minority, but I use Windows 8. I run a 64-bit version of Cygwin. All the standard Gnu software is already compiled and is only a click away. I used to dualboot with Linux, but it became tiresome, and I like OneNote.
Also, GCC software like Cloudy compiles and runs just as fast on comparable Mac or Linux hardware. Now that Apple no longer supplies gcc with XCODE and their llvm compiler is frequently broken (Cloudy issues as of 2014), I find the Cygwin option to be better.
Also, when I spent 3 years on a 2009 Mac Pro I found GCC software, compiled with the same version of GCC, ran 20% faster in a Linux VM inside OSX than it ran natively. That is an odd performance hit.
Thanks Jane for the info. I just think that the title+conclusion induce bad interpretation; astronomers use — as the chart shows — four of the major operating systems on earth.
Particularly, I can not define the best one; and I think because it doesn’t have one. The hard work is done (more and more) inside Linux boxes (no discution as far as I can see): take a look at the supercomputers list operating system family http://i.top500.org/stats to see who does the crunch. Daily work, yes, Mac is taking a lot of space, and I don’t dare to say why, but I can tell you my example.
I use both, without preference. The cluster I administer and crunch the numbers: Linux, for sure (CentOS) — Stable, simple, light… linux way of life. My laptop, where I do my daily work (web/mails, programming, studies/research) while listening to music (let alone Starcraft ;): MacOSX (10.7) — BSD, great GUI, “smooth” use.
Though I think Linux is much better at the low-level (tools for administer and develop on the command line), MacOS *X* version was greatly implemented with the Darwin+Aqua design, which brings the most from both worlds (MS/Windows & GNU/Linux), with performance and stability.
Linux, in good flavors (Slackware, Debian, Gentoo, RedHat), is great; it just works, easy, for the hard work. MacOS is great for the daily/home work, where one can taste and test our data and techniques.
It would be interesting to see the numbers from the European side. Something tells me it won’t be as “mac-heavy”
Actually this would be an interesting exercise for me to do looking at how this differs for the Multi-mission Archive at Space Telescope.
This is a laptop/desktop issue, not a question of machines used for science processing of whatever sort. Obviously users will submit proposals from the machines they use for text processing (what a quaint phrase) and web browsing (becoming quaint itself). There is a clear growing preference for Macs, but why assume these are high-end systems and not the very competitive iMacs? Once you assemble a Linux desktop with a similar quality display, etc, the price advantage (even ignoring the self-administration issue) is not really that great. At least as significant a trend here is the plummeting of Sun, once king of the astronomy hill. There’s no sign that the Windows and Linux markets are collapsing, rather that personal Mac systems continue to do well. The Mac server market is nothing to write home about. Which is to say that it is misleading to place the Macs at the top and the three downward trends at the bottom. Remake the figure just swapping Sun to the top position and the Windows/Mac discrepancy won’t be artificially over-emphasized. Better yet, remove Sun entirely – I don’t see any HPs or SGIs confusing the plot. It’s perhaps more interesting sociologically that a couple of proposers appear to still be hanging on with ancient Suns.
Bring back VMS.
It would be nice to have the same statistics on what OS is used for successful submission ! Then we’d know who are the brightest (just kidding!).
In common with several previous comments, I would be counted under “Mac” since my Cycle 19 submission when in from my laptop, but i’m primarily a “Linux” desktop user. In a similar vein, I suspect most of the APT “Windows” users will be those using Windoze laptops but other platforms for desktops.
Two comments, one question.
Comment: It might help to remind some of the participants that OS X runs BSD Unix under the hood. Therefore, the “Linux vs. OS X” debate is a bit like arguing that a Pontiac Firebird is better than a Chevy Camaro; same engine and transmission, different body cladding and upholstery.
Comment: Ben’s comment in #3 is thoughtful and worth re-reading. Go re-read it. I’ll wait.
One question: Could y’all please flesh out for me the assumption that proposals are submitted from laptops rather than desktops? Why? Do people agree with this assumption? Speaking for myself, I submit proposals from my desktop unless I’m traveling.
Regarding the comment, Linux and BSD (which are kernels) share virtually nothing in fact, besides broadly following the Unix interfaces (POSIX). Mac OS X does not even use the BSD kernel (which is equivalent to Linux). However it ships with BSD tools (sed, etc.). I ran into quite a few problems because they tend to have incompatible options with the GNU ones shipped in linux distributions. So the body cladding and the upholstery are indeed different, but the engine and the transmission are too.
Um, what?
What Ann said. I used FreeBSD exclusively for many years, and only went Mac when I bought a laptop (definitely easier, with all the proprietary hardware). I’ve since switched to Mac as a desktop too, since it’s easier just to use one system, but the ONLY reason I ever went to Mac is because it’s based on FreeBSD.
Nothing against Linux, I’ve used it too… but I’m fond of BSD.
more data points for public archives are available here: http://archive.stsci.edu/surveyresults/index.html
Interesting stats (even if they always have to be taken with the necessary caution).
Without being too obvious about the promotion of my own software (already presented in this blog: https://www.astrobetter.com/iobserve-the-astronomical-observing-app-weve-been-waiting-for/ ), I will agree with the “happiness” criterium mentionned above. For whatever good or bad reaons, I like to work on Mac, even for doing hard computations. I feel more productive. I considered my Mac as a Linux box for work, with all the benefits of having a finished and well-crafted machine (leaving aside shiny-but-sometimes-crappy-software like iTunes & so on).
All Linux people finding hard to install open-source software on a Mac are probably giving up a bit too quickly.
As one data point for you, Jane, I also use my desktop preferentially as long as I’m in my office.
The question of “which OS is it easier to install software on” is only slightly more interesting and relevant than “which is cheaper” (a question I agree is just a distraction down in the noise). I am a big fan of the integrated package managers available on most Linux distros – but you can get near the same experience on a Mac using macports these days. There are also bulk installers like the Scisoft CDs and Enthought Python Distribution for both OSes.
Personally, I used to years ago sometimes spend days wrestling with installing software, but those times seem to be long past for me. But I gather that’s not true for all software of astronomical interest? Maybe that’s a poll for another day, “which astro software packages are the biggest pains to install (and how do we fix that)?”
Oh, man — ever tried to install BOA? Some kind of bolometer analysis package (clearly I don’t use it myself). I used to build my own Linux kernels, and BOA nearly broke me.
I use linux pretty much exclusively, and to me the integrated package system is a huge plus. I very frequently have to help grad students who generally use a mac laptop. Even with the help of the local sysadmins and macports I have to say that it is really a pain. On the other hand my linux distribution of choice already has quite a few astronomy packages which are perfectly integrated into the system, and I made some extra ones (which is pretty easy), so installation and update on my work/home systems is very easy (basically, it is nearly automatic). And I know I am not the only one who uses these astronomy packages as I got feedback about them. The only packages I install manually are IDL (which I hardly use at all anymore), and IRAF (which is very slowly improving but still feels very 1980s).
In the end I have the feeling that many people who had a negative experience either had it 1) quite a few years ago when linux was arguably not as user friendly, and/or 2) the system was maintained by a sysadmin. Regarding point 1, I have been using linux for a decade now and the progress over the last few years has been most impressive. Even with a laptop it just works nowadays. Regarding point 2, I also had very negative experiences about that, with sysadmins installing obsolete versions, being reluctant to update and being not too responsive to install the required software. Anywhere I work now I find an agreement with the local sysadmin: I do not bother him about anything and he lets me do whatever I want. In the end everything works and I must spend at most a few hours a year doing administration (including making new packages).
Our department runs nearly exclusively Macs. The reason is that about six years aego our system administrator left back to Canada. This meant someone had to step in and be sysadmin as we were mostly linux machines. I “volunteered” and after about a year of this temporary situation I got the department to agree that Macs would be the only computer I would support and we would phase out all the linux boxes. We operate all Macs now and I’m still the sysadmin and I like the direction we went. I when new people arrive that they generally accept that we are Mac and are happy to fall in line as they don’t want to be responsible for software installs, etc. I understand that linux is the main cluster operating system. This makes sense. They actually pay people to look after those clusters. So you might argue about costs of the machines but look at our institution. We are saving the cost of a sysadmin (I’m not happy about that but look at from a business point of view) which in New York the total cost would be upwards of 150k per year. We spend less then 10% of that yearly. Anyway that’s my two cents.
As another data point, I do all submissions from my desktop (which these days is a Linux box, unfortunately).
What I find absolutely fascinating in this discussion is how individuals can have such radically different impressions of the same OS. Several commenters have mentioned how difficult it is to install anything on Mac whereas Linux is a breeze…..which is precisely the opposite of what I would have reported! I’ve been using Macs for over five years now and I’ve always loved how simple installing software is. My new Linux box that I have to use at work, however….ugh! I’m far less productive with it and I’ve yet to successfully install anything (though I’ll admit that this is most likely heavily biased because I am not allowed su access).
Anyway, it’s just interesting to me how varied our experiences are. Marshall’s point is a good one for discussion: which software packages are the most difficult to install (and how does that differ across OS’s). Cheers.
I wonder how much of this trend is actually driven by mobility, given how much astronomers travel. Does the APT store information beyond just operating system that might be used to determine if the submitter is actually on a laptop or a desktop? As some of the other posters have suggested, Macs definitely have a more compelling advantage in the laptop market – I switched from a linux laptop not too long ago, and am much happier with a macbook pro… lots of things just work more consistently and cleanly when I’m traveling, the battery life is much better, while pretty much all of the linux utilities I’m used to also work fine on a Mac. Most of the people I know who have switched have similar motivations: “It works better for new stuff, and the old stuff all works just as well.”
If Apple ends up doing what Ben and Peter suggested, and phase out “computer”-like laptops or just make OS X less and less compatible with linux stuff, I imagine we’ll see this trend reverse. It really doesn’t make sense to be doing our kind of day-to-day work on an iPad…
I use a Mac because it is fast or me to use. Brain-time costs far more than CPU time these days. I have colleagues who are Linux-users from way back and can’t imagine living their lives any other way, guys who share DNA with Linux code — and I can’t help noticing that they spend a lot more time than I do on sys-admin tasks that my Mac takes care of automatically. Those guys definitely have greater capability than I do, and can accomplish things I could never dream of. However, I got into this racket to do planetary science, not computer science, so I am comfortable with my choices.
I have never owned an Apple product. Because we astronomers are more tech-savvy than the average person, I’ve occasionally been asked to help friends or relatives (and higher-order variants, like relatives of friends) with their MacBooks. I guess it’s partly for lack of familiarity but I find them infuriating. I just feel like I can’t do what I want to do.
We use Linux boxes in the department. My major problem there is that our sysadmins, who are otherwise very good, still have us running Red Hat 5. So here I am browsing using Firefox 3.6. That shouldn’t be a mark against Linux, though. It’s just our admins. I’m very happy dual-booting Windows and Ubuntu on both my laptops.
One valid complaint about Linux that hasn’t been mentioned is the lack of support for some hardware. When I got my first 10″ Asus Eee PC in late 2008, getting the Wifi working under Ubuntu 8.10 was a bit of a mission. That Eee was stolen and replaced it at the start of this year. At that point, everything ran out of the box on Ubuntu 10.10, except that Asus remove the only hardware Wifi switch, so I had to install an applet to let me enable/disable Wifi because the Fn+F2 option still isn’t supported in Ubuntu. The situation is improving but I confess that it’s possibly still an issue for some, especially with laptops. Desktop hardware is broadly better supported.
Ultimately, I’m a very happy (and daresay proud) Linux user. I’ve never seen the appeal in Macs/OS X.
I use Macs, but I have good friends who use Linux, and they still seem to be pretty nice people….
Seriously, this discussion has the same flavor as the VMS–UNIX war of 20 or 30 years ago, or vi vs emacs, which still goes on at some level today. It seems to be hard to discuss such things without a judgmental tone creeping in.
When I switched from Linux to Macs 10 years ago, the main draw was being able to use Word, Powerpoint, and Excel without having to reboot into Windows. I imagine things are different now in the Linux world, but I have gotten used to Macs, and I don’t see any point in changing back unless Apple completely gives up on providing suitable hardware or makes MacOS too inflexible (I have to say that I still do miss having true focus-follows-cursor; it is available within X-window, and MondoMouse helps somewhat within Aqua, but nothing works when going between X and Aqua). And I do like Pages as a cleaner alternative to Word (yes, I do use TeX or LaTeX for most things) and I like Keynote better than PowerPoint. But the point is: this is not a moral issue; it is simply what one is used to and comfortable with. Occasionally I am asked to help someone with a Windows PC, and I always feel like I am all thumbs, and I get very frustrated. Perhaps it is similar (though I would guess to a lesser degree) when a Linux user tries to use a Mac or vice-versa.
In this area, it seems that familiarity, instead of breeding contempt, breeds sectarian fervor.
I’m currently involved in a packaging project with Mageia to try to make it the distribution of choice for astrophysicists.
The problem with Linux is that the distributions that are aimed at astrophysicists don’t have the polish and easy of use that Mac OSX has. It’s not surprising that Linux is losing market share to Mac OSX. If you are just using a computer for ordinary things, then there is no reason to use any of the Linux distributions out there.
There’s only one problem with the Mac and that’s that you are limited to what Apple wants you to do . Now for 99% of the stuff that you want to do with the computer, that’s enough, but there are things that can be done with the 1% in which you want/need total control over the hardware.
What I’m trying to do is put professional tools on a linux distribution. I ended up with Mageia because it has the basic infrastructure to be a good solid distribution. Now Linux is *never* going to be as easy to use as a Mac. But If I can get Linux to be 80% of ease of use of the Mac, and then put “must have” tools so that astronomy geeks are willing to live with the 20% of annoyance, that’s good enough to be useful. The nice thing about Linux is that I can make it better without getting the permission of Apple.
If anyone is interested in helping just reply to this comment. If nothing else, I’d like to get a list of packages that people *need*. I’ve got ds9 working, and I’m going to get IRAF working.
Very interesting project joequant! I am not an astronomer (at least now as a college student) but that’s definitely an interesting project. I used to be a Mandriva user and now I’m working on Archlinux. Mandriva has almost died due to their bad commercial decisions but there’s Mageia now.
But I do think that even for everyday jobs, Linux is much better than Mac. You could have Awesome or Xmonad as the windows manager, mutt as the email client. They do have a learning curve but once I am familiar with that, I found that they are much better than anything like KDE/GNOME or Mac OS X. Even if you are doing presentation, it’s not complex to use python scripts like landslide to convert a lightweight markup file to a HTML5 presentation. Asciidoc is even more convenient, it’s easy and fast to write, it supports math formulas and it could generate the markup into HTML, PDF, LaTeX, Epub and HTML5 slides. Inkscape is also a good tool for doing presentation. With the plugin Sozi you could easily create Zoom UI presentations based on SVG and javascript. Inkscape also has a good support to LaTex. These tools are much more efficient than both Keynote, PowerPoint and LibreOffice Impress. And you could make the presentation minimalism but also in pretty cool style, which could be much cooler than Keynot and PowerPoint. With these simple tools, I didn’t see any reason to use Keynot, PowerPoint or LibreOffice. But I do agree that Linux has many low-quality applications like Kmail. But in each area there have been some good applications.
As Archlinux is not stable enough, I am considering switching to Gentoo during the summer and focus on using XMonad.
I believe that Linux is truly losing market in the astronomy circle. But in the whole world, Linux is still in uprising together with Mac OS X while Windows is in declining.
For natural science, I believe that Linux is essential due to a fact that many great computing clusters and supercomputers are running Linux. Some are running BSD. That’s Linux’s advantage. But for personal use, it’s true that many would prefer Mac OS X although Linux is more efficient. I think a key factor is that today many college students are using Mac products. If they are using Windows, they might switch to Linux after studying in astronomy. But if they are using Mac from the beginning, they might not switch as IRAF and other tools are also available to their platform. That’s why Mac is so successful now.
I love Linux but I confess that many astronomy software is like a hell. It’s hard to install. In fact in Linux I could also install them easily with Scisoft packages, but they would simply mess my system up. Many of these software is not designed for the advanced package managers of most Linux distributions. That might also be a factor that many are using Mac now. As the software designers look to be not familiar with tools like URpmi, pacman, apt-get and emerge.
I think what Linux could do better is to have overlays that could simplify the installation of language-specific packages. Because I do prefer Haskell program, this is a huge problem. And in Mac OS X there’s a fatal disadvantage that you are forced to you cabal. In ArchLinux I could package myself but that’s just a waste of time. In Gentoo I could use the Haskell Overlay. It’s simple and convenient.
Overall, it’s a very interesting project! I do agree that using Mac means loosing control. In China there’s a rumor that Apple upload almost every iphone’s contact book to the Internet.
I’m a Linux and Mac sysadmin at an 8-meter observatory. These stats line up pretty well with our science user base. Without getting into a dogmatic debate over the altruistic merits of Mac vs. Linux (I prefer both, frankly), I can tell you pragmatically that most of our staff tend toward Mac OS because of MS Office support, our MS Exchange mail & calendaring, and the slick integration of an easy and predictable GUI on top of a powerful BASH or TCSH shell environment. MS Exchange access from Linux plainly sucks. I have no control over our mail infrastructure, and I frankly feel like the choice of Exchange is an odd one considering our platform demographics, but alas, it is what it is — I wasn’t here when the Windows admins were hired…
Our production telescope systems are all Linux, as are hundreds of our servers, but all but 3-4 of our scientists have at least one Mac. I have a soft spot for OS X, as I have never once had to deal with hardware drivers. Linux has gotten quite a bit better in this regard, but Linux on a laptop still has the worst return on investment in terms of support overhead, manageability, and reliability compared to every other platform in our organization.
Have a look at Distro Astro, a Linux-based operating system for astronomers. You can read about its features here: http://www.distroastro.org/features
MS Windows 7 with a 64-bit Fortran-77/90/95 compiler from the Portland Group could be considered a serious number crunching machine. That’s the setup I use. My system has the Titan NVidia GPU, and when used with Portland High-Performance compilers, can be considered a serious contender for doing numerical computations (over 2 TFLOPs). This setup can do some serious work such as stellar modeling, nbody calculations, and other professional astronomical codes that can be compiled using modern Fortran or C++. Why can’t Windows 7 64-bit operating system be used more by professional astronomers? You can run IDL, Mathematica (which can use the GPU), different languages, and the easy to use Microsoft Word+MathType for writing thesis, papers, etc. It seems that a lot of heavy-duty COTS software that can be used by the astronomical community is developed for MS Windows 7 in contrast to either Linux or the Mac. Is it simply that scientists don’t wan’t to be caught using an operating system that has received a bad name, or is it more of a “macho” attitude of using “real” operating systems like Linux, that require hacking experience? I would think the minor cost difference in software between free Linux software and relatively cheap COTS software under MS Windows would not be a deciding factor on which operating system to use. I only wish IRAF/STSDAS could be ported to MS Windows, but I don’t see anyway to do this, and there are no alternatives to IRAF/STSDAS (other than home developed IDL code under MS Windows). The only spectral analysis package I could find is “SpecView” developed at the Space Telescope Science Institute, which allows basic spectral analysis.
Hi Mark,
I don’t see any reason why you couldn’t use Windows. My hypothesis on the popularity of Unix-ish tools (Solaris, Linux, Mac OS X) is due to the fact that so many folks in astronomy learned to hack FORTRAN, C-shell, Python, C, and Perl while they were neck-deep in their respective academic programs. Yes, you can shoehorn these environments onto a Windows box, but they come natively on most Unix platforms with zero extra effort.
For my environment, Linux is the platform that actually controls our telescopes, so having Linux as a production environment just makes sense — nothing machismo about it. And the choice of going with Linux vs Windows was not made as a religious or fanatical choice, but purely from a pragmatic stance that Linux can be stripped away to it utmost bare essentials for the role it is serving, where as Windows, Mac OS, and even Solaris can only be stripped down to a small degree without destabilizing the platforms. I think you’ll find this is a strong motivator behind using Linux as a number cruncher — it can be stripped down to such a tiny footprint with almost no overhead at all. Less resources for the OS = more resources for computation. In the end, the right tool for the job is the one that enables you to be the most productive. There is no one-size-fits-all.
Cheers,
Chris
Seems I replied to early in this thread. I won’t rehash but I think anyone on Windows should be looking at the newly finished 64-bit version of Cygwin. It performs fantastically, and you get all the gnu software, terminal support of Linux or Mac in Windows. And now that apple has removed support for things like GCC and X11, anyone needing those on Windows or a Mac has to got through the same steps to get them.