Re: statistics graphs
I think Jean is right to rename We also might rename the plot_boxplot plugin which implements all these plots to plot_stats. However, all the univariate distribution graphs are perhaps named something like plot_stats_dist since there are 2D and 3D graphs, as well as some of the high dimensional graphs (castles, glyphs, faces, etc.) that eventually may get implemented. It may be worth looking at the R names, not necessarily because they are a great choice, but in case we want to borrow ideas/setup etc. One trap for the unwary: boxplots and histograms etc. on small samples can be very misleading and give different results with different packages. This is the issue of how to divide 3 chocolate bars among 4 kids. I've some references if needed. Cheers, JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: bad file
I've put a sanitized file at macnash.admin.uottawa.ca/files/badfile.zip I didn't try this on my laptop, which has suffered enough today, but did try loading in a Virtualbox instance of Ubuntu Hardy with Gnumeric 1.8.2 (the Hardy default) installed. This actually loaded the file, but I then had to "Force quit" Gnumeric. Tried twice to be sure. So it has at least some nasties that a likely worth investigation. Cheers, JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: File causes lockup
Will do what I can. The names occur multiple time on different sheets, including in the sheet names, so it is many, many times to sort out. Unfortunately, we've been sent explicit instructions that linking names and student numbers in files that are shared with non-authorized staff is a no-no under Ontario privacy legislation. JN Jody Goldberg wrote: On Sat, Jun 07, 2008 at 12:50:46PM -0400, Prof John Nash wrote: Some teaching assistants sent me a file related to assignment marks that has a sheet per student (about 60 sheets), each with graphics. This will not load in Gnumeric 1.8.2 that comes with Ubuntu Hardy. There is lots of disk activity for over 5 mins, and no keyboard or mouse control of machine. Even Ctrl-Alt-Bsp and Ctrl-Alt-Del gave no response. I eventually had to hit the power. Confirmed with 2 tries. OO and kspread seem OK, as does Excel 2007 in WinXP. Sounds like we're over allocating memory. Any chance of fuzzing the student names in a different application and sending of a copy ? ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
File causes lockup
Some teaching assistants sent me a file related to assignment marks that has a sheet per student (about 60 sheets), each with graphics. This will not load in Gnumeric 1.8.2 that comes with Ubuntu Hardy. There is lots of disk activity for over 5 mins, and no keyboard or mouse control of machine. Even Ctrl-Alt-Bsp and Ctrl-Alt-Del gave no response. I eventually had to hit the power. Confirmed with 2 tries. OO and kspread seem OK, as does Excel 2007 in WinXP. Are there known issues with 1.8.2 file open? Because of marks, I cannot post the file. However, I am likely meeting Jody for some other work in a few weeks if this turns out to be a genuine issue. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: summer of code projects
This is very good news. In case it is useful (and perhaps others will add notes), I've put in some comments on related work: Mariusz Adamski : 3D Plots A physics senior from Wroclaw University of Technology in Poland who will be working with Jean to add surfaces. -- Duncan Murdoch (Maths and Stats, U. of Western Ontario) has done quite a lot of work on the R package RGL. It does rotating 3D graphs. He's approachable and helpful. I've just done a little item on printing very large graphs with him that I put up on the r-project.org wiki. R and Gnumeric have a long and friendly history. Daniel Hall : Audit Trails A CPA studying CS at the University of Alabama at Birmingham. He will be working with Morten to extend the undo/redo code to provide persistence. -- Daniel already has been in touch with me, since the telltable.com / telltable-s.sf.net projects were a motivation to this (but I'm hoping Gnumeric will make them redundant!) Capturing history is not too difficult, capturing so it can be used easily and cleanly is, I think, likely to be quite hard. There are some commercial packages out there (System 7 and Wimmer Systems come to mind) that seem to be able to ask for very big bucks from the pharmaceutical industry for clinical trial data handling and from financial institutions for handling their investment analysis models etc. David Torne Berga : Multi-Dimensional Data Visualization Will be working with me on 'data slicers' (aka pivots and pilots) -- There's a hard core of Excel users who claim this is the reason they can never use anything else. I believe in one close-to-home case it is the justification for spending on 100s of copies of Excel, though possibly only one user ever uses them -- the person controlling the purchase! Best wishes to those working on these projects. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
More on jhbuild scripts
The default jhbuild modulesets try to get gcrypt elements from a gnupg site that no longer is functional. (Stage 2 etc of the build.) Nor are individual source codes there, but one can download a large bz2 or gz tarball. Perhaps there's one or two others out there interested in fixing the scripts with me. Doing it alone takes far too long, while back and forth gets things fixed better and more quickly. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: build issues
Following advice from Priit Laes, I got svn working again but had also got it going on a new virtual Debian. Unfortunately, I seem to be getting loads of bugs with jhbuild, particularly concerning python-related problems. This afternoon, I even tried blowing away jhbuild and re-installing as per live.gnome.org/jhbuild/, but even that is halting with an error claiming I have only automake1.8 rather than 1.9, and failing to find COPYING or INSTALL. Are others seeing similar problems? Or is it time to check my hardware? JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
jhbuild svn issue
I've spent some time trying to downgrade subversion from version
1.4.6dfsg1-4 without success. Apt was VERY persistent in keeping that
version. Unfortunately, while I can checkout, for example, gnome-common
using
svn co http://svn.gnome.org/gnome-common/trunk gnome-common
jhbuild gets "Network socket initialization failure". On the web under
debian bugs I find Peter Samuelson commenting on having such a problem
and someone else saying they would look into the jhbuild script
('jhbuild update' fails too).
I'd really rather not start from scratch to rewrite something like
jhbuild. This looks like some sort of jhbuild / debian mismatch that
could be nasty to resolve.
JN
___
gnumeric-list mailing list
gnumeric-list@gnome.org
http://mail.gnome.org/mailman/listinfo/gnumeric-list
svn checkout woes
Seems there is a bug report bugs.debian.org 480038 The problem is in the libsvn1 library on client machine. I'll try to figure out the fix. It affects a good deal of jhbuild stuff. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: gnumeric subversion repository query
I'm using anonymous access to my knowledge. And steps 1-4 go OK. I've just checked that I can run svn co http://svn.gnome.org/svn/gnome-common/trunk gnome-common and get an updated version. So the issue is more likely somewhere in the scripts (again!). JN Adrian Custer wrote: > Any chance this is due to your ssh key being thrown out? Sounds like the > right level in the interaction. If you haven't heard, all debian derived > distros going back a couple of years were generating trivially weak keys > so everyone is resetting their ssh keys. > > --adrian > > > On Tue, 2008-05-20 at 22:14 -0400, Prof J C Nash wrote: > >> In retesting my build process, I'm getting >> "svn: Network socket initialization failed" >> at step 5/47 of the build (gnome-common). >> >> This looks like something is wrong with the subversion server rather >> than at my end. Am I correct? If not, any hints where to start debugging? >> >> Cheers, JN >> >> ___ >> gnumeric-list mailing list >> gnumeric-list@gnome.org >> http://mail.gnome.org/mailman/listinfo/gnumeric-list >> ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: gnumeric subversion repository query
In retesting my build process, I'm getting "svn: Network socket initialization failed" at step 5/47 of the build (gnome-common). This looks like something is wrong with the subversion server rather than at my end. Am I correct? If not, any hints where to start debugging? Cheers, JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: IEEE 754 compliance
The complexities of the "edge effects" are best kept out of the spreadsheet, as Morten indicates. However, there are some computations that might be influenced by how a particular internal calculation is performed. I was earlier looking at the ends of the Gaussian (normal) distribution where one gets some weirdness in Excel. This could be because very small numbers are handled poorly. It is in the special functions etc. that I would think Gnumeric and other spreadsheets are most likely to be interested in IEEE 754 and its revision. Must admit I was unaware of 754r activity. Thanks Dave. For info, there's a nice Wikipedia item at http://en.wikipedia.org/wiki/IEEE_754r with a link to a recent essay by Velvel Kahan (who was really the person who got all the floating point stuff going, and in the 1960s pretty well embarassed IBM into retrofitting the 360 with guard digits for floating point) that gives a nice and nasty example using Excel. I've tried this in Gnumeric and get somewhat different results, but which I'm sure would upset novice users. JN Morten Welinder wrote: > Gnumeric does not let you access NaN etc. It would interfere with the > desired semantics. > > Morten > ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: IEEE 754 compliance
There are some resources for testing by Nelson Beebe at http://www.math.utah.edu/~beebe/software/ieee/timops.html From what I understand, the main issues are handling of edge effects (underflow, overflow, divide by 0, etc.) where compilers may do some things different from the standard's prescription. There is, of course, an interaction here with hardware that may not provide ways to get at the bits (literally). In my efforts to set up some Gnumeric test worksheets, I've tried to contact Beebe without success. He may have retired (I believe he is older than I, and I'm on the brink of retirement from teaching, but not from Gnumeric!) If there is interest, and in particular an example where IEEE754 may be important, I'll be happy to dig a bit. I was a corresponding (ie vote by mail) member of the IEEE 754 committee back in late 70s. Given the arcane detail, it will take a bit of review for me to get fully up to speed. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
jhbuild / libgsf more info.
In case it helps others: I managed to get a gnumeric build by downloading libgsf version 1.14.6 and doing a "manual" build of it that matched jhbuild, namely - download from http://download.gnome.org/sources/libgsf/1.14/libgsf-1.14.6.tar.gz to ~/checkout/gnome2/ and unpack - enter the libgsf-1.14.6 directory - './configure --prefix=/opt/gnome2' so the build goes to the same place as jhbuild's efforts - 'jhbuild buildone gnumeric' (I'd already run everything else with 'jhbuild build gnumeric' before getting an error that the libgsf was 1.14.5). Hopefully, I can gradually get more proficient in working with the scripts and provide information, fixes, and possibly a build-bot on one of my research servers. One of the first things will be to get jhbuild up to date with gnumeric. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: jhbuild / libgsf
"gnumeric.spec.in is not maintained, see bug 159782. However, 1.14.6 is correct Morten" The problem, then, is that jhbuild structure (inc. latest modulesets) is out of date, as it builds libgsf 1.14.5. I'm prepared to learn how to fix such issues and to dedicate some time to maintaining and documenting such matters on an ongoing basis. Indeed, I come from a management school and have identified this as a (the?) major threat to open source projects. However, I can get there a lot quicker -- and save developers dealing with msgs on this list related to build problems -- if I get some input from someone with expertise and experience with the scripts. Anyone? JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Another jhbuild libgsf glitch
Having returned from an academic visit to Perth WA, I'm getting back to my attempt to set up a build-bot to allow for building of versions of gnumeric that are up to date e.g., Windows cross-compile. I've still lots to learn. My current problem concerns a matter that has appeared several times in the archives, namely libgsf. The jhbuild default for building libgsf creates 1.14.4, but the gnumeric.spec.in wants 1.14.6. Can someone point me to documentation on how to put in the correct settings? The JHBuild manual gives the syntax, but not the location of the defaults. There's also some glitchiness with scripts for gtk+: Here there appear to be no "make" rules for install all or test all, but ignoring the errors seems to be OK. However, I'm ultimately going to want to avoid stoppages and manual intervention on these scripts. For information, my goal is to have a web page to select the settings for the build and send an email notification when the result is done. And I'm painfully aware that this will mean figuring out how to package up everything so it works for other folk. Collaboration welcome -- this sort of thing is not part of my computer background in numerical and web programming. My gut feeling is that in order to get the system I want set up, I'm likely to have to document and clean up some of the scripts, which will be useful anyway. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: JHBuild problem in tester for gtk-doc?
Is anyone else seeing an error in tester.c in the build process for Gnumeric when gtk-doc is being built? I'm getting what looks to be a pretty straightforward typo (an unexpected parenthesis error) when running JHBuild. If this is a "known" issue, I'll just wait a couple of days. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: 1) 600K rows 2) Windows compiles, 3) missing values in graphs
For both the long rows and the missing values in graphs, I think R might be a relatively quick solution. R can read tabular data in CSV form and similar, and stores things in "data frames" that are like spreadsheets but with column type. And missing values are supported for many if not most of the capabilities. www.r-project.org. Note that Gnumeric and R have a long cooperative history together. Compilations: I've been exchanging ideas with Jody about using a server I have to allow him to run builds. I've "sort of" got it running and installed jhbuild etc., no-password ssh logins, and tried a test build (failed because I'm missing a file or two, but getting there). Since Jody has plenty on his plate, if there's anyone with some experience running builds who wants to work with me to set up appropriate nice features, please get in touch off-list. I hope to set up a small set of web documentation that could answer some of the periodic questions on how to build and possibly offer scripts or other resources bundled from working builds. The server is a VM running Debian lenny. At the moment the host hardware is only at about 10% CPU tops, even supporting about 20 VMs. As long as there's no interference with those other VM applications, the U of Ottawa will be happy to allow a useful activity such as builds to be done. Apparently the VMWare scheduler is set up so we likely cannot be too greedy, though I propose to show proper etiquette and be watchful of resources used. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: Gnumeric dies on two-means, unequal variances.
In Gnumeric 1.7.8 distributed with Ubuntu 7.04 (Feisty), selecting the Statistical Analysis tool, two-means, unequal variances is like doing Alt-F4 or Ctrl-Q: Gnumeric closes. Is this a known bug? If not, I'll submit a report. J Nash ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: Polynomial regression - Warning
Use of the condition number is sensible. If one can detect polynomial regression, however, there are alternatives that may improve the condition, and it would be nice to tell folk. That's not necessarily Gnumeric's job, of course. And there may be more important "bad habits" that should be dealt with first. My experience is that users of polynomial regression don't want to know they are doing something silly, and are wilfully blind to the issues. There's of course the annoying problem that condition numbers are not easy to estimate Another story. I'll hopefully be able to provide some input and reasonable (i.e., plug out, plug in) fixes after a few more months of wading through other projects. JN Morten Welinder wrote: > Unfortunately we cannot really know if the user's use of the tool is sane. > Yes, a chain saw can be dangerous but it is fundamentally a good tool > when used right. > > What we can do -- at least for the GUI regression tool -- is to check the > condition number of the relevant matrix. We do that already and warn > when the result starts to lose significance. And err when it is completely > gone. > > Morten > ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: Polynomial regression - Warning
The type of dialog Adrian suggests would be a very sensible feature. Sometimes one really does need to do the polynomial regression, even if to show the issues, so it is not right to completely bar an approach. However, when we have an interested person, it's a good opportunity to point them towards potentially better methods. There was some discussion of such dialogs in the 80s for statistical software, and there's some of these ideas in some of the packages. Certainly I'm interested in helping out on this sort of thing, and have been gradually clawing away at other stuff I have to do so I can return to the test spreadsheets, which are one way to bring the issues to light. JN Adrian Custer wrote: > Lovely repartee, just the sophisticated answer that gnumeric brings to > the spreadsheet world. > > Any chance you can craft this into a good popup dialog? e.g. > > You are trying to use SOME_METHOD which exists in gnumeric only > to allow compatibility with other older spreadsheet programs. > The computations involved in that method are known to be > exceedingly problematic. However, you might be able to solve > your problem using SOME_OTHER_METHOD. Would you like to use that > instead? > > Of course, this is all over my head. I'm merely hoping to leverage your > 40 years of knowledge into a helpful dialog. > > --adrian > > > On Tue, 2007-10-09 at 14:46 -0400, Prof J C Nash wrote: > >> Polynomial regression was where I came into numerical analysis at the >> beginning of my academic career 40 years ago. It is a dangerously >> ill-conditioned problem, meaning that the regression parameters are >> untrustworthy, though the "fit", i.e., the model, may be useful if the >> calculations are done properly. There are ways of doing it that are >> "less dangerous" using orthogonal polynomials. >> >> Just a warning that it should not be a high priority to add. Let the >> Excel users drive off the cliff. If it is added for Excel compatibility, >> then I'd still recommend a "Gnumeric does Vista" and pop up warnings >> asking if the dangerous move should be allowed. Come to think of it, one >> could borrow exactly the Vista popups so folk would blame Bill. >> >> Seriously, think if you really need polynomial regression or can use a >> less troublesome approach. >> >> JN >> >> ___ >> gnumeric-list mailing list >> gnumeric-list@gnome.org >> http://mail.gnome.org/mailman/listinfo/gnumeric-list >> ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Polynomial regression - Warning
Polynomial regression was where I came into numerical analysis at the beginning of my academic career 40 years ago. It is a dangerously ill-conditioned problem, meaning that the regression parameters are untrustworthy, though the "fit", i.e., the model, may be useful if the calculations are done properly. There are ways of doing it that are "less dangerous" using orthogonal polynomials. Just a warning that it should not be a high priority to add. Let the Excel users drive off the cliff. If it is added for Excel compatibility, then I'd still recommend a "Gnumeric does Vista" and pop up warnings asking if the dangerous move should be allowed. Come to think of it, one could borrow exactly the Vista popups so folk would blame Bill. Seriously, think if you really need polynomial regression or can use a less troublesome approach. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Query regarding hypergeometric function
This is a query about what "should happen" when a function is given
arguments that are in some sense out of range.
The hypergeometric distribution function computes the probability that x
"successes" are observed in a sample of n trials from a finite
population of N elements where M of the N are "successes. We sample
WITHOUT replacement. Example: 10 people in a village of 100 have math
anxiety. We sample 5 of them and clearly don't check the same people
twice ('without replacement'). Hypergeometric distribution gives the
probabilities we get 0, 1, 2, 3, 4 or 5 with MA in our sample.
Now suppose we have nobody with the disease. Clearly P(0, 5, 0, 100) =1
and the rest should be 0. Gnumeric reports for "=hypgeomdist(1,5,0,100)"
the result
#NUM!
which has some merit, since we are doing something that is "impossible"
(getting 1 out of none).
I can see reporting an error for
=hypgeomdist(1,15,0,10)
i.e., sample bigger than population. But I know I'd rather get 0 for the
values that are "impossible" in a feasible sample.
There are several of these borderline cases throughout the functions. It
would be nice to document them and then come up with a good set of choices.
Comments welcome.
JN
___
gnumeric-list mailing list
gnumeric-list@gnome.org
http://mail.gnome.org/mailman/listinfo/gnumeric-list
Virtual build environment
I have made this a separate posting from my build woes as it may be of value to the Gnumeric and other communities. As I had some difficulties doing the Gnumeric build in Ubuntu, which I now believe to be due to the Ubuntu folk having to change a set of library file names because of a name collision between two pieces of software, I have been building in a virtual machine. At first I used VM Ware server free version, but I believe that may be time limited. Now I"m using Innotek's VirtualBox and have found it quite useful, indeed in more ways than just for builds. I have tested copying the VBox "hard drive" of my virtual machine. It's a little big at 5.1GB, but when I got it on a second machine it came up and had preserved the state -- my final error msg was on the screen in the terminal window. It seems to me that a VBox "hard drive" set up to to builds would be a relatively painless way to get more folk able to do the builds. Setting up JHBUILD is non-trivial. I plan to post some notes when (if?) I succeed. However, I know I would have jumped at doing a bittorrent download of a disk image. Am I a late-comer to this idea? It seems a no-brainer except for the file size, which is big, but not exceptionally so. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: Continuing build problems with JHBUILD
Following up Jody's msg (removing files from checkout to ensure cairo build etc.) allowed my build to continue to the last stage 47/47. But ... Now I get an error that appears to be something omitted from the script. Here is the output: /bin/sh ../../libtool --tag=CC --mode=link gcc -g -O2 -DG_DISABLE_DEPRECATED -DPANGO_DISABLE_DEPRECATED -DGDK_PIXBUF_DISABLE_DEPRECATED -DGDK_DISABLE_DEPRECATED -DGDK_MULTIHEAD_SAFE -DLIBGLADE_DISABLE_DEPRECATED -Wall -Wmissing-prototypes -Wsign-compare -Wpointer-arith -Wnested-externs -Wchar-subscripts -Wwrite-strings -Wdeclaration-after-statement -Wnested-externs -Wmissing-noreturn -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations -Wno-pointer-sign -module -avoid-version -L /opt/gnome2/lib -lpython2.4 -o python_loader.la -rpath /opt/gnome2/lib/gnumeric/1.7.12/plugins/python-loader python-loader.lo gnm-python.lo gnm-py-interpreter.lo py-gnumeric.lo py-interpreter-selector.lo py-command-line.lo py-console.lo boot.lo -lm gcc -shared .libs/python-loader.o .libs/gnm-python.o .libs/gnm-py-interpreter.o .libs/py-gnumeric.o .libs/py-interpreter-selector.o .libs/py-command-line.o .libs/py-console.o .libs/boot.o -L/home/john/checkout/gnome2/gnumeric/plugins/python-loader -lpython2.4 -lm -Wl,-soname -Wl,python_loader.so -o .libs/python_loader.so /usr/bin/ld: cannot find -lpython2.4 collect2: ld returned 1 exit status make[3]: *** [python_loader.la] Error 1 make[3]: Leaving directory `/home/john/checkout/gnome2/gnumeric/plugins/python-loader' make[2]: *** [all-recursive] Error 1 make[2]: Leaving directory `/home/john/checkout/gnome2/gnumeric/plugins' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/home/john/checkout/gnome2/gnumeric' make: *** [all] Error 2 *** error during stage build of gnumeric: ## Error running make *** [47/47] [1] rerun stage build [2] ignore error and continue to install [3] give up on module [4] start shell [5] go to stage force_checkout [6] go to stage configure choice: Suggestions welcome. I seem to be so close! JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: gtk+ build problem - cairo-xlib.h
Since having some pretty severe problems a year ago with building the latest gnumeric under Ubuntu due (apparently) to library naming issues in that distribution, I've had some success by running the build under Debian Testing via a virtual machine. First I used VM Ware, but after changing to new hardware, I installed Debian Testing under VirtualBox since there is an Ubuntu package for this. However, running JHBUILD to build Gnumeric in this environment, I encounter an error at stage 18/47 (gtk+). The error is given as gtkdrawable-x11.c:32:24: error: cairo-xlib.h: No such file or directory I've tried installing some of the dev libraries such as xlibs-static-dev, but no joy. Also, since cairo-xlib.h is available in the src collections elsewhere, I've tried placing it in /opt/gnome2/include/cairo/ but that then produces an error "cairo was not compiled with support for xlib backend". I suspect that there is a minor bug in one of the configure or build scripts. Anyone have suggestions for overcoming the problem? JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: Median and other quantiles
The rather heated discussion on the median has a historic quality. It is not quite as old as the arguments about how many angels can sit on the head of a pin, but not too far off. Perhaps I can cool things down by providing some perspective. There are, to the eyes of someone who has been teaching and working with this stuff for nearly 4 decades, three issues that are causing conflict: 1) The question of what we actually want "MEDIAN" to compute. 2) The question of how to express this clearly. 3) The question of how to actually do it. Question (1) concerns the allowed values for the quantity returned by a function of type "quantile", the general term for a median, percentile, or any other quantity that returns the "value in the scale of the data with rank k from the minimum (or, alternatively, from the maximum". We can insist on returning values from within the original data set, in which case we must sometimes choose between two values, or we can specify a rule for interpolation. The most common rule for the median is to interpolate when the number of elements, n, is even, leading to the "definition" that the median is the average of the middle ranked values for this case, while it takes the value of the middle rank datum when n is odd. I believe all the spreadsheets use this rule. Percentiles are usually (but not always) computed using interpolation, and I have seen some variety in the values given for quartiles in different statistical packages. Rob Hyndman published a quite good paper in The American Statistician about 8-10 years ago on the subject. John Tukey (stem and leaf graphs, box and whisker plots, five number summaries, Exploratory Data Analysis, etc.) liked to have measures that belonged to the data set. So he defined Depth as the rank of an observation from the nearest "end" (highest or lowest) and Tukey boxplots use Hinges rather than Quartiles (which are generally interpolated, but not always by consistent formulas). See http://macnash.admin.uottawa.ca/files/stembox.txt for a bit of a discussion on Depths etc. Ultimately, this question boils down to where to cut to divide 4 candies among 5 children. No matter what you do, things get ugly. For large n, whether you use interpolation, and what formula for interpolating, becomes less important. When we come to (2), I am surprised nobody is using RANK rather than SORT. To compute quantiles, one does need to rank the data in some way, but we don't actually need to sort it. There were a number of articles about 30 years ago on ranking vs sorting, and if anyone is interested I can try to dig some up. If I were writing the MEDIAN definition for the TC, I would change it in one word to: MEDIAN logically _ranks_ the numbers (lowest to highest). If given an odd number of values, MEDIAN returns the middle value. If given an even number of values, MEDIAN returns the arithmetic average of the two middle values. In other words, I'll live with the interpolation. Actually, I believe Tukey (someone correct me if I'm wrong, my copy of EDA is in another office) defined median as the value with maximum depth from either end or the arithmetic average of the two values having maximal depth if the value is not unique. Question (3) is one for the programmers and algorithm performance people. I think a fast ranking algorithm would be very useful to improve performance in a lot of spreadsheet features, not just the median. But it should be an algorithm that is quite "clean" so that application to more than just a column or row range doesn't end up causing all kinds of potential for bugs. For the sake of development and testing, it may be useful to use compile or even execution switches based on some stored control parameter, so that informed users can try out different ranking algorithms. Note that this may be helpful for increasing sort performance on very large sheets. Note that RANK() gives the rank of a number in a set. We actually want the number that is ranked at some level. No doubt the discussion will continue. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: interest rates
The important issue is what conventions are used for time and rate. About 25 years ago I tried to get information on this from Canadian banks. Some were cooperative. As I recall, the three that responded used three DIFFERENT rules. This was for weekly payment mortgages. In Canada, there is a little known law that prescribes that mortgage interest be computed "annually or semi-annually, not in advance". Moreover, if nominal annual rate is over 6%, the borrower must be provided with a schedule of payments, or everything defaults to the 6% rate after the fact. There've been some interesting commercial mortgage cases from the early 80s where rates were around 20% and the schedule was not correct. To get to monthly, weekly or daily mortgage payments, you have to know how many periods there are. For monthly payments, we can use 6 months, so the "working" rate is 100 * [(1 + nominal_rate/200)^(1/6) - 1 ] Is everybody still there? Weekly or daily? Well, there are, as I recall between 181 and 185 days (I should check this, it's been a while) in a "half-year, depending on the start date and whether one uses calendar "date". Or using days, one has to decide when things end. Or 365/2 = 182/5. But 182.5/7 is a bit more than 26 weeks, worse in leap years. Which is where the fun begins. I'm not sure I want to put canned formulas into Gnumeric or any other spreadsheet for this, and I would definitely like to see more transparent output for mortgage payments. Of course, in the UK (or at least England as Scotland may have its own rules) most mortgages are demand loans so use floating rates. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: strings in gnumeric / awk / etc.
Some of the issues being raised suggest that a spreadsheet is not the right analytic tool. How about a data frame in R? There are easy transformations from spreadsheet to data-frame and back (and they should be better set up but are not to my knowledge!). R allows character strings to be converted to factors which can be very useful in some analyses e.g., regressions. There are quite a few character handling functions, and some support for regexes, though I will not claim any expertise in using the latter. My own opinion, as a supporter of gnumeric, is that we sometimes try too hard to do everything in the spreadsheet. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Bug Buddy failure!
I managed to crash Gnumeric 1.7.0 (on Ubuntu Edgy) three times trying same operation. Had one spreadsheet (xls, opened in Gnumeric), added a sheet, copied a region from another file (wb3, opened in Gnumeric), tried paste. Bang. On third try, I'd opened a new spreadsheet and copied the wb3 info there first, then copied again. So then I thought I'd use Bug Buddy. Not so good! "Bug Buddy has encountered an error while submitting your report to the Bugzilla server. Details of the error are included below. The component specified doesn't exist or has been renamed. Please upgrade to the latest version." For the record, things work in ooffice 2. Maybe this is known. If so, I'll try to upgrade asap, though it's easier to use Ubuntu package updates. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: accuracy ...
The discussion is revisiting a lot of work that was done in the 1950s and 1960s on floating point arithmetic. The gold standard is to accumulate in a double length mantissa. On some architectures that was easy to do. The i386 architecture uses IEEE 754 as far as I am aware (I have not looked at details for a while). This makes it fairly easy to double length accumulate reals, but almost all current s/w uses doubles by default (roughly 15 decimal digits equivalent). This means accumulations have to be done in quad, which wasn't part of the standard toolkit. If I ever get jhbuild to complete, I may even be willing to try. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: gnumeric-list Digest, Vol 27, Issue 1
I've seen a number of msgs on this list about compiling a verion, usually the latest, of Gnumeric. I've tried myself too. What I haven't found yet, and not only for Gnumeric, is a good how-to on actually doing such compiling safely and cleanly without disrupting one's working environment. That is, I have 2 "main" working machines, one running Xandros 3.02, which gets Gnumeric 1.4.3 (sigh!) using apt-get, and Ubuntu Dapper, which get's (as far as I can tell) 1.59 (it may have just got to 1.61). But to really check tests, I need the latest release. The difficulties are that if one tries to install the "new" libraries, existing and needed applications can become unworkable. Some sort of safe sandbox (chroot environment) is likely needed. But some good, sane, advice would be helpful. I've had suggestions of putting up Gentoo, but I've played distro-roulette enough to now be very careful about making changes. I've also some critical, everyday things I must keep working. To ensure that I'm not just complaining, if someone sends me rough notes and I get things working, I'll be happy to edit and prepare the HowTo and to the extent my schedule allows maintain it. I use Linux, but I'm prepared to try to help out on other platforms with editing a HowTo or possibly running a WinXP boot that I do have available. JN (nashjc _AT_ uottawa.ca for off-list communications) -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: compiling gnumeric etc.
As Morten points out: > It is not simple and there are more traps than you might want. My own experience is that it is not too difficult to GET the code, but the configure/compile/link are far from trivial. The main issues in my case have been libraries that are not as recent as Gnumeric wants. Unfortunately, installing newer libraries can "break" other applications. There are ways round this I've heard about but not tried, in large measure because I haven't found a good tutorial that might be entitled "How to safely build a program that uses more recent libraries than your system uses". I'd be happy to learn of such a HowTo, or to contribute to its development by being a suitable test "stooge". One thing I have done is use Debian "apt-get source gnumeric" to get a version of the source code I COULD build. In my case I managed fine with 1.4.3 on a Xandros 3.02 box, and 1.5.9 on Ubuntu Breezy. JN -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
test of normal distribution functions
Morten is right that some values input for inverses tests are not fair to any spreadsheet processor. I hope to find ways to provide more informative output, that is, how the calculation fares in relation to what is reasonable at the default precision. The current example is VERY preliminary, and I simply copied values from one test to another. Some of the more interesting tests are the bounds ones. These allow comparison with simple functions that can be locally computed. There are other examples of such functions, and a compromise is needed between different objectives to get a "nice" test. For the moment I am avoiding the r.foo functions. Later on I hope to add to the Gnumeric-specific test files. For the moment I'm concentrating on ones that can be saved as .xls. I find these useful to send to Excel addicts. JN -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
New test spreadsheet for normal distribution
Much, much later than I had intended, I have prepared a more detailed test of the normal distribution functions (normdist, norminv, normsdist, normsinv). This is still not complete, but shows some directions that tests of these functions could take. The file is at http://macnash.admin.uottawa.ca/files/normtest.xls There is a "rubbish" Sheet3 which can be ignored. I welcome comments, suggestions, and offers of collaboration on this work, which I intend to gradually extend to other functions. JN -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: Attempt to provide warning of sorting bug
After a bit of struggling, I managed to download and compile sane collections of code to experiment with. Morten suggested looking in commands.c, where I found the sort command quite easily and the call to sheet_range_contains_region function which is, I believe, the check for arrays in the sort range (in sheet.c). I think I need to use sheet_foreach_cell_in_range for a first check, but am wondering if there is any write-up, even provisional, that will give me some examples of use. Afraid I am someone who learns mostly from examples rather than reading the book. I've been around long enough to know an example that works is worth a lot more than a book that may have errors. Essentially I need to learn how information is passed between the sub-programs, as it appears the setup is pretty organized with common data structures to move info around. Plan is to first figure out if I can properly detect different conditions, namely: 1) is there a cell reference within the sort range If there is, then 2) a) which of these references are absolute refs OUTSIDE the range (OK) b) which are references within a row (sorting rows, so refs are across columns) (should be OK) c) which are references to other cells, possibly in the sort range (not OK) Once I have code to test these conditions, I'll need to learn more about the GUI, as I want to be able to highlight cells, particularly the dangerous ones. For info, a colleague had group work in his class and was marking the teams. If column F contained the marks, and row 1 was leader of team 1, row 2 team 2, row 3 team 3 and the rest of the rows were the rest of the students . He entered the team mark, then used =f1 or f2 or f3 to apply the mark. Then sorted! A mess. If anyone else thinks this a worthwhile warning to add, do get in touch outside the list. Looks doable without too much pain. I'd like to make it work so folk can suggest how to do it better before suggesting it as a new feature. JN -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: sorting with forumulas or references
Finally managed to get bugzilla to send me a pw. and found discussion related to formulas in sort ranges as bug 59144. It appears to remain an open issue. JN -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: sorting "bug"
Whether this is a bug or a feature or ?? is not clear. What is clear is that it is TROUBLE. My colleague Andy Adler sent me a msg pointing out that spreadsheets (OO, Gnumerice, Excel) all take the sheet A B 1 3 1 2 2 =B1 (<- set to 1) 3 1 3 and sort it on A to A B 1 1 3 2 2 =B1 (<- set to 3) 3 3 1 It is also interesting to sort on B. Longer examples show that the cell with the reference can be moved to where it makes no sense and gives a #REF error. For example: A B 1 6 2 7 3 6 = B1 4 9 and sort A in reverse order. Gives an error as =B1 points to =B0 (doesn't exist). The sorting is taking the data in the cells and ignoring the reference. The big danger is when we don't get any indication that we've made a reference, but the data is changed, as in marks for 600 students in my statistics for business courses. Nasty! What should we be doing? 1) formulas with references INSIDE the sort area should not be sorted. Indeed, one could say that sort should pre-check this and red-flag it. As far as I know (I'd be delighted if there is a way to do this), no spreadsheet does this yet. 2) formulas with references OUTSIDE the sort area should be OK. 3) references across a row for row sorts or down a column for column sorts should be OK, but we should warn the user that the references are there. (yellow flag) Burns in his Spreadsheet Addiction paper http://www.burns-stat.com/pages/Tutor/spreadsheet_addiction.html warns of this danger but in more general terms. Is this worth putting into the bugs system? Or is it a feature request? Or ??? J Nash -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: www.gnumeric.org?
I'm wondering if there is any progress on this. I've several sets of slides on the web that I'll need to update if it keeps going to the Novell site. JN -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: Formatting rules -- one use?
It seems that the formatting could be used in interesting ways when following an audit trail. Jody and I have had discussions on this, and perhaps he can enlighten the group on status of the audit trail capability. This would be useful to me in particular as I am giving a technical talk to the Ottawa Carleton Linux User Group about spreadsheets and Gnumeric especially (Nov 1) and a somewhat more broad talk on spreadsheet addiction and its risks for government and business to some relatively senior federal govt. folk on Nov 18. Most will not have ever heard of Gnumeric, so I'm hoping to get at least a little play in that direction. I'm also trying to improve the function test spreadsheets. Contributions and discussions on these efforts welcome. JN -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: handheld Gnumeric
Since there have been some items here on Gnumeric for cell-phones etc., the call for papers on HandHeld computing at http://old-www.cis.uoguelph.ca/~qmahmoud/sac2006-hhc/ may be of interest. JN John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list [EMAIL PROTECTED] http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: missing "help"
Having persuaded a friend to start using Gnumeric on her first Linux PC, I am embarassed that she points out that the Help functionality is not present. I don't have her version of Libranet distro (a Debian variant), but my own Xandros 3.0.2 let 1.4.2 be installed via apt and it also says it cannot find the "specified file" when I ask for help contents. I am willing to help find and document this bug, but could use some pointers as to the "usual" place to find the help file. If it turns out to simply be mis-located, perhaps we can get someone who knows "deb" packages to assist in fixing. John Nash, U of Ottawa School of Management ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re: directly executable Gnumeric (was USB/CD version)
I apparently failed to make clear that my wish is to have an executable of Gnumeric (and other Gnome stuff too) that can be run from CD or USB. I've remastered Knoppix once (see www.telltable.com), but that is a very different situation since one reboots. I'm looking to plug in a USB and run Gnumeric on a "foreign" system without installing it. Some Win apps and some Linux ones can be run this way, but I'm not familiar with the requirements for so doing. Truthfully, there could be obstacles that render this idea infeasible, but clearly it could make it much easier to "show and tell" and also to get folk trying good applications. JN ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Gnumeric from USB key or mini-CD?
This message is both a query and a suggestion. I am wondering if anyone has run Gnumeric from a USB key or from a CD. If so, under what platforms? My feeling is that one of the biggest obstacles to the wider use and adoption of Gnumeric is that folk have to install it. If we had Gnumeric on a USB key and/or on a CD (mini CD with logo would be nice, or even full size Gnome tools), especially if major platforms were present, those of us who use it could run it when we need to, even as guests on other machines. I'm not an expert compilation/configuration, but willing to work on this as I think the eventual payback would be large. I'm guessing that lots of folk would use Gnumeric / AbiWord etc. if they didn't have to install it. My own particular interest is when I need to run ssconvert somewhere to extract files from .xls. Comments on feasibility and desirability welcome. JN -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
Re:ssconvert - from gnumeric-list Digest, Vol 14, Issue 26
I'd like to add to comments re: ssconvert. I've found it useful for doing a number of things like converting between spreadsheet formats. Also in doing a large number of experiments on forecasting, I output 1 "row" of tab-delimited items per run to a file (I append to a text file, which is pretty easy in just about all applications), then use ssconvert to make an .xls file. (No flames please, my clients have clients who don't "get it" yet. Sigh.) One complaint: I so far (in Win XP -- again, not by choice, as things run fine from wherever in Xandros Linux) seem to have to move files to the right C:\Program Files\gnumeric\bin\ directory before running ssconvert because otherwise it cannot find the appropriate dll's. Does anyone know if ssconvert could be recompiled to include all the dll stuff? Or some other mechanism so it is in the PATH appropriately and can be run from any location? Cheers, JN -- John C. Nash, School of Management, University of Ottawa, Vanier Hall 451, 136 Jean-Jacques Lussier Private, P.O. Box 450, Stn A, Ottawa, Ontario, K1N 6N5 Canada email: nashjc on mail server uottawa.ca, voice mail: 613 562 5800 X 4796 fax 613 562 5164, Web URL = http://macnash.admin.uottawa.ca "Practical Forecasting for Managers" web site is at http://www.arnoldpublishers.com/support/nash/ ___ gnumeric-list mailing list gnumeric-list@gnome.org http://mail.gnome.org/mailman/listinfo/gnumeric-list
