Slashdot Mirror
Study Recommends Gnumeric Over MS Excel
Jody Goldberg writes "A recent study of analytic quality, and responsiveness to problems strongly preferred Gnumeric in place of MS Excel. With new problems popping up in Office XP the case for spreadsheet users to migrate is only getting stronger.
In some related Gnumeric quickies, a new stable version 1.2.6 was released, and Open has done an interview with the Maintainer."
In a recent interview, Jody said a W32 port was the priority. I think that could actually start pushing it over the top and make some real headway, I can see why it would be a priority.
"Elmo knows where you live!" - The Simpsons
The most recent release of Gnumeric is amazing. The only downside is that I often have trouble deciding between OOcalc or Gnumeric, and often flip back and forth depending on my mood. Both are worthy competitors for Excel.
I love that one of the "failings" of Gnumeric was that the random number generator function RND was *too* random - Gnumeric uses the /dev/urandom device that generates random numbers from noise sources in the system (noise diodes, interrupt events, user input, etc.) rather than using a psuedo-random number generator with a predictable sequence.
True, there are times it is nice to have a "random" number generator that you can re-run for testing, but having a really random number generator is better for a host of problems.
www.eFax.com are spammers
It seems as if they're moving way too prematurely on this. In the article it said they posted to a Microsoft newsgroup and didn't receive a reply, and that this means that Microsoft will never fix the bug. Obviously there -may- be some tech support people roaming the newsgroups, but it would've made much more sense to simply contact Microsoft's technical support department and talked with someone directly about this error.
This is similar to having your car found defective, and then placing a flyer downtown to ask the company to contact you about options instead of picking up the phone and dialing the correct number.
I'm not a fan of Microsoftian ideals, but wouldn't that have made more sense before going all this way?
I use excel at work all day and I have to say that no Open source solution comes close to providing what I expect a spreadsheet to do.
The idea that one should switch from excel to an open source solution because of a small set of statistics problems cannot be properly solved by excel seems a bit like throwing the baby out with the bath water. (unless you do nothing but statistical modelling all day)
It looks like Gnumeric improved or stayed the same on every data point except Pidigits, Numacc2, and Origin1 (whatever those are). Note that the LRE is the negative of the log of a value less than one, so a larger LRE means a smaller relative error. It's just the number of digits that agree with the correct answer. Really bad values would even have a negative LRE.
The ocean parts and the meteors come down
Laid out in amber, baby.
Someone had to pay for this little Gnumeric study! Get out the torches ... oh, wait.
The reason people won't switch away from MS Excel has nothing to do with technical specs and everything to do with the very large number of Macro's and templates already written. There is an awful huge installed base for whom Excel works fine, and they don't see the problem. Most of the financial services sector for example. From there point of view, it's not broke Why fix it?
If TODAY everything was equal, there would still be a 10 year lag until a change happened, as that is the roll out time, and the time to convince people they 'want' to change. It better have some kick butt feature that they don't have in Excel, or they are going to resist change. That is just the way people are
Gavin Fischer
I've e-mailed a well-informed and helpful Microsoft developer, whom I first encountered on this very forum, on several occasions. I'm told a number of bug reports have been filed against the application in question as a result of my e-mails, and some of the things I've mentioned to him have certainly been fixed in a later version of the product.
Some people at Microsoft do listen, you just have to make a bit of an effort to find them. Curiously, a comment from the developer in question was that the dev teams love direct contact with customers prepared to give them helpful information about bugs or feature requests, they just wish the PR people would stop getting in the way. :-)
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.
>[...] a larger LRE means a smaller relative error
Precisely.
>It looks like Gnumeric improved or stayed the same on every data point
So far so good...
> except Pidigits, Numacc2, and Origin1[...]
... where the new version scored worse according to the table; lower numbers meaning fewer correct digits.
Even so the description reads:
"As can be seen, Gnumeric 0.67 used an unstable algorithm for computing the sample standard deviation, and on this basis its performance can be considered unacceptable. This was fixed in Gnumeric 1.1.2." (my emphasis)
This to me is inconsistent, but I'm still open to the possibility that I'm misunderstanding something.
Belief is the currency of delusion.
Keep in mind that inertia works both ways. Yes, there's a lot of folks that don't want to move. However, it's equally difficult (possibly worse) for Microsoft to regain any customers that do move. Also, actual movement tends to lag decision-making for a while, so visible market share lags actual inertia by some amount.
Finally, keep in mind that even upgrading from one version of Excel to another can break compatibility. The office world has very strong backwards-compatibility requirements. Gnumeric may not fill those requirements, but we also know that Excel doesn't do so.
May we never see th
The easiest way to get new versions of Gnome software is via garnome!
Garnome 0.30.1 was just released and it features the latest version of Gnome (2.5.5), The new, non ugly file dialog (but not all programs use it yet) and of course, Gnome Applications, including Gnumeric 1.2.6.
It is designed for IA-32 Gnu/Linux, but it should work on most OS's. Download it now.
And if you liked the power of garnome, you may be interested in the power of Gentoo Linux, which is like garnome for your entire distribution!
found here
/dev/urandom is not /dev/random. When the entropy pool is exhausted (which will happen extremely quickly if producing a large set of random numbers for statistical work), instead of blocking it will use a hash algorithm. /dev/urandom varies unpredictably between being unpredictable and being very unpredictable.
:-)
On the other hand, when doing a study, frequently you *do* want to be able to use the same seed to produce exactly the same results. This is a legitimate failing in gnumeric. Not all random numbers are created equal.
For what it's worth, I did (simple) analysis of a large set of random number generators for a high school science fair project. The Microsoft RNG (which has been used ever since at least early QBASIC days) is pretty decent, at least from a uniformity standpoint.
May we never see th
I've seen spreadsheets where MS-Excel would miscalculate results by 20%. MS-Excel also has enormous problems handling circular spreadsheets. Both are probably related to defects in the order-of-calculation algorithm.
Can we please not capitalize "the Maintainer"? It's a bit too "Logan's Run" for me, or for anyone I think. Geekness has overstepped boundaries when Those Who Are In Control of Software are afforded the same nobility in print as the King, the President, and the Messiah. Let's remember that software engineering is a discipline, a job, and that we, as a group, can't produce bugless office software, much less achieve such status in society that we must be addressed by our titles, that those titles must be honored, that the masses must gaze upon us and tremble ... which, by and large and not incidentally, they do---but only geeks would assume that it's in awe of our deep knowledge of C++ and Java ...
Chr0m0Dr0m!C
I didn't read all of the linked article -- so whatever...however, I will say this: Anything that makes Microsoft Office look bad and (insert cheaper solution here) look better, I like.
For a $1000 computer, I pay ~$400 per license for MS Office Professional -- that's 40% of the cost of the computer. If I could convince management and our user base, I'd change to anything else because anything else would be cheaper (Star Office, Lotus Smart Suite, OpenOffice, whatever). I checked out Open Office with one of our accounting guys, and it worked just fine with all of his macros. Peace of mind against FUD just isn't worth that much. MS Office is a fine product, just not worth the price. If there was anything with a remotely competitive amount of market share, I'm sure that MS would drop their prices to stay competitive.
-Turkey
Fixed in later versions means the customer had to buy a new version to get the bug fixes. That isn't support that is sales.
The only thing I use Excel for is Solver. Solver turns Excel into the worlds easiest to use linear/non-linear optimizer for ANY function you can put in a spreadsheet. I use Gnumeric a lot, but I always have to go back to Excel for Solver...