This is a RFE.

Will continue work on this isssue asap.

Just to add my 2 Brazilian Centavos, I feel that this is quite possibly the
single most important feature (that doesn't currently exist in StarOffice) for
the scientific community in terms of usability of Opencalc.  I know a number of
people that are using Excel *only* b/c this feature does not exist in SO.

Is there anything I can do to help?  Testing?  Example data? Non-gui coding?

changing QA contact from bugs@ to issues@

Would be nice to have some feedback on the status of this issue...

*** Issue 3444 has been marked as a duplicate of this issue. ***

A nice-to-have but not on our roadmap for 2.0
Set to OOo Later

> A nice-to-have but not on our roadmap for 2.0
> Set to OOo Later

This decision torpedoes the chance of OOo being adopted as site policy
in the very many higher education establisments where (non
mathematical) scientific departments exist. There has been plenty of
support for this feature. Since Excel Chart contains this feature, end
users will resist adoption of OOo as an office suite, no matter how
good the rest of it is.

Don't get me wrong, I use Gnuplot for charting, and probably would
even if the ability to set error ranges were included, but the
majority around me will cling to Excel Chart for dear life if the
alternative lacks what they see as a basic feature.

OOo could realy use the education users for medium to long term
growth, and it frustrates me incredibly that the developers appear not
to be listening to the users on this...

I saw somewhere that the ability to define flexible source range
selection for different parts of data series is in the roadmap for OOo
chart 2.0 (http://tools.openoffice.org/releases/q-concept.sxw) - is
this element so alien that error ranges can't be done with the other
ranges?

I have to second Derek's comments.  The ability to get a quick graph
of data (including error bars!) is really fundamental for OOo adoption
in the scientific world.

Well, as always, keep up the good work people.  Not complaining, just
wishing.

Falko,  
it is not a nice to have, it is essential for scientific use. And that is exactly were 
OO.org/StarOffice can find most potential users. 
I tried to hack it in there myself, but understanding the internals of OO.org sadly 
costs me to much time. 
 
Also, it is a bit strange that you post "continue this issue asap" in 2001, and 
postpone it 2 years later to somewhere after 2.0. So the status "Started" seems 
silly to me. 
 
Yes, I am complaining :) 
 
d. 
 

I too agree that this is a more important issue for the adoption of
OOo Calc by scientist.  I started using OOo Calc as my spreadsheet,
but had to always open Excel in the end to make figures for my
cumulative data. Sadly, I now only use OOo Calc if I'm pretty sure I
won't need such a graph. I agree with Jon and Danny that this feature
is very important if you want to attracted scientists-- a group who
may be more willing to make a switch that those in e.g., business.

gregg 

*** Issue 20551 has been marked as a duplicate of this issue. ***

Glad to find others to agree with that this issue is an essential fix
and not a "we may possibly fix this later" type of issue.

I was very supprised to find that OOo could not do this as it is such
an important function for graph drawing. Especially when publically
presenting data, without error bars your results can be viewed with
some suspicion. 

I have to agree with Jon Lapham that you are blocking any wider
adoption of OOo, especially by schools, universites and the scientific
community by not fixing this issue ASAP. It's a shame you changed your
minds from "will continue work on this isssue asap" in 2001 to "A
nice-to-have but not on our roadmap for 2.0" in 2003.

However, OOo is a great product that is always improving so please
keep up the good work!

Re-assigned to Matthias Müller-Prove for further evaluation.

to Bettina, our central dispatcher for RFEs

Anybody wanna figure out how to do it?  Surely the OO people would
help us or point us in the right direction.  I know C and a little C++.  

Also I gotta say- I agree with everyone in this thread, and would love
to ditch excel once and for all, but come on--it's free.  Funny how
the more professional the package begins to look the more people seem
to feel they've somehow paid money for it.

It might never work, but if there were a few of us in contact, playing
with it at our leisure we might actually overcome that steep learning
curve.

Imagine the fame.  Imagine the glory.  Imagine all the women falling
at our feet.  OH how they would fall.

Jack
jackinthepants@businessfist.com

Anybody wanna figure out how to do it?  Surely the OO people would
help us or point us in the right direction.  I know C and a little C++.  

Also I gotta say- I agree with everyone in this thread, and would love
to ditch excel once and for all, but come on--it's free.  Funny how
the more professional the package begins to look the more people seem
to feel they've somehow paid money for it.

It might never work, but if there were a few of us in contact, playing
with it at our leisure we might actually overcome that steep learning
curve.

Imagine the fame.  Imagine the glory.  Imagine all the women falling
at our feet.  OH how they would fall.

Jack
jack@businessfist.com

*** Issue 21749 has been marked as a duplicate of this issue. ***

*** Issue 22861 has been marked as a duplicate of this issue. ***

*** Issue 22740 has been marked as a duplicate of this issue. ***

*** Issue 22740 has been marked as a duplicate of this issue. ***

*** Issue 23909 has been marked as a duplicate of this issue. ***

*** Issue 22429 has been marked as a duplicate of this issue. ***

*** Issue 25277 has been marked as a duplicate of this issue. ***

I don't believe that this IS fixed as I've checked the user manuals from 5.2 to
now and find the same instructions for plotting trendlines and it simply doesn't
work.  Oh yeah, why do we have an option for a "power" regression where we can't
set the power that we want to use.  I believe 1 of 2 things happened. 
StarDivision was working on this before being purchased and it has remained the
same since then or whomever was working on it just stopped before the acquisition.

Here's a detailed example of how it DOESN'T work.  Like everyone else here, this
is not a feature or enhancement but a BUG and a NECESSITY!  Nobody at Purdue
even looks at this because of this "feature".  

Here's the data that I'd like to perform a regression analysis on:
0	0
500	250
1000	400
1500	460
2200	495
3000	512
3600	520
4000	525
4500	520
5000	515
5500	500
5900	490
6500	460

I create an x-y plot with the data which looks fine.  Then I double-click on the
line (within the plot) and receive a window that essentially has the "object
properties" such as: line, characters, font effects, data labels, statistics and
options.  Under options I should be able to select the regression curve of
choice; again, please note that the power variable of "what power is desired" is
not there.  Anyhow, if any of the regression curves other than linear are chosen
such as log, exponent, or power are chosen then no proper curve appears on the
plot.  In fact, I get a line across the bottom of the plot that is supposed to
represent my "power regression".  This doesn't work according to all the reading
that I've done and I've done a LOT.  

  The lack of having a polynomial fit, moving average, or even something as
simple as an R^2 analysis in the regression package is also pretty laughable. 
How are businesses, students, governments, and academia supposed to use this
when they can't EASILY massage their data.  Just look at Excel and after
plotting the data, then click on the line, right-click and left-click on "add a
trend line" and you'll see what you are missing.  It sounds like the OO guys
that are supposed to be working on this don't appreciate (or don't
understand/comprehend) how this is a VERY BASIC functionality when dealing with
statistics.  It is an ESSENTIAL part of BASIC statistics and it is MISSING.

It's apparent that this isn't going to get remedied by the chaps at OO in the
near future as long as they keep overlooking it.  Well, I'll just keep saying
that OOcalc is a joke in comparison to Excel.  Nice job with the other packages
but if you want to see how it is supposed to work then look at Excel.

I know my way around numerical methods and being able to solve for these
parameters so I can help with the coding if anyone else wants to figure out how
to get OO to work with it.  If we wait on these guys to release it in VERSION 3
AT BEST then we'll be old and grey.

Otherwise, if you want this item resolved by the OO guys, like I do, then "vote"
for it by logging in and casting some votes.  I think we are allowed 5 by
default to cast on OOcalc.

sorry if I'm curt but I've been waiting forever and the fact that this gets
passed up everytime someone brings it up is frustrating.

Find this a defect and not an enhancement.

*** Issue 28458 has been marked as a duplicate of this issue. ***

This is a very important issue. Graphing is often overlooked, but it is a
critical tool that must be flexable enough to allow the user to add and change
graphs to produce what they need. I am a Biologist and I need to be able to
create graphs that have error bars that are based on 'real' values contained in
a range of cells. For any kind of graphing I must use another program (currently
Applix for Linux). I agree with many of the other comments here, that this is a
BASIC feature that must be present for this to be adopted by scientists,
students and educators. 

Hello Björn, wiil the error bars be integrated into OO.o 2.0 including ranges? 

My report, 28458, has been closed as a duplicate of this one, but I'm not now sure.

I was suggesting that it ought to be possible to access the regression
coefficients in the sheet, and more importantly, that it ought to be possible to
do so without drawing a graph at all!

What do other reporters feel about this?

Yup, I'd like to see that too.  I'm also SICK of seeing this as an
"enhancement".  What good is a spreadsheet if it can't do basic regression/error
analysis easily and in a user-friendly way.

There is basic regression functionality in Calc. With the functions
SLOPE(yrange,xrange) and INTERCEPT(yrange,xrange) you get a linear regression
f(x) = SLOPE*x + INTERCEPT. You can also calculate r^2 with the function
RSQ(yrange,xrange).

All other regressions in the chart are based upon linear regression and can be
calculated easily with the logarithm:

logarithmic and power: create a column with the LOG(x) of every x-value and use
this as y-values
exponential and power: create a column with the LOG(Y) of every y-value and use
this as y-values.  Instead of INTERCEPT you have to use EXP(INTERCEPT)

Hope, that helps.

BTW. about regression curve parameters there also exists Issue 5289.

Sorry, I meant:

logarithmic and power: create a column with the LOG(x) of every x-value and use
this as x-values (not y-values!)
exponential and power: create a column with the LOG(Y) of every y-value and use
this as y-values.  Instead of INTERCEPT you have to use EXP(INTERCEPT)

@bobharvey,enmane,bm:
Have a look at the fuctions LINEST() & LOGEST() to calculate various parameters,
inluding the regression coefficient, RSQ,... and others 

setting keywords, subcomponent according to new RFE-process

*** Issue 40578 has been marked as a duplicate of this issue. ***

I've just downloaded OpenOffice.org 1.9.74 for SuSE, and I was really surprised
that this functionality is still lacking. Will it be available in the final
release of OO 2.0?

I looked for information on this functionality and the best I found is a macro
called FitOO, which is discussed in the French forum of OpenOffice.org.
The macro is available at http://oooconv.free.fr/ However, it's not as easy as
in Excel. 

As already said by many other users, this is not an enhancement, but the lack of
an important feature.
My work is based on data analysis, and I don't want to create columns or use any
other workaround if I can do that with a single click on Excel.
I don't understand why it's so difficult to add such a functionality. Calc
already plots the regression curve, so it calculates its coefficients. 
Few lines of code will add the capability to show the equation in a text box.
Few lines of code which would make Calc more useful for a lot of people in the
scientific world!
I consider this feature a *lot more important* than the new logo for
OpenOffice.org 2.0, which could be the same of 1.1. with just the release number
changed!

Regards,
ap

OK, enough is ENOUGH!  This issue needs another evaluation.  It shouldn't be
under "graphics" and is DEFINITELY not an "enhancement".  This

There are a whole slew of people that refuse to use this product because of the
lack of this functionality.

This is addressed to bm, cloph, joergwartenberg, and mmp.  Whos IS it that you
report to?  Surely everyone reports to someone else and there HAS to be someone
that can reevaluate whether this is an "enhancement" or a SERIOUSLY LACKING
FEATURE.  I would like have you guys/girls reevaluate as to whether this is an
enhancement or not.

Maybe we need to educate you guys/girls on what is lacking as maybe you are
unaware as to how important these features are to ANYONE that has to ANY
statistical analysis.  Children in High School are exposed to this using Excel
and almost every college student in ANY scientific field uses these features
REGULARLY.

For some, maybe spreadsheets are a way to tabulate data and run simple equations
but for the scientific community it is used for this and MUCH more when we need
to evaluate the data statistically.  If these features cannot or WILL NOT be
added then you are FORCING us to continue to use Excel.

Who's attention do we have to get to have them take this more seriously?  

To everyone else that finds this a problem then use your VOTES to escalate this
issue. 

I for one agree that the "pretty" features and the database issues are moot
compared to this feature.  I ask myself the following question all the time,
"who doesn't use the statistical features of a spreadsheet?"  For this reason
alone I find it completely incomprehensible that these features are STILL not
there.  This product will remain second-rate until these features are included.

So either you are not getting it or we aren't doing a better job explaining what
the problem is.

I second the last comments, this is a severely lacking feature that prevent the
basic statistical uses of the spreadsheet, mostly with non-technic savvy people.

The usage of macros or the LINEST funcions are not enough, as most students want
to simply put data in a column and see the regression line with its regression
coeficients, at least.

Please reconsider this bug, and the closely-related ones in Issue 5289 and Issue
4509.

Thanks in advance.

Last time I looked at the relevant source code a placeholder had been put in the
code by one of the developers for custom-range defined error bars, so it looks
like the intention is there, although perhaps not in time for 2.0.0 itself (this
was several months ago, though). Perhaps someone who has the time at the moment
could have a look and report back.


Another important point. Some recent comments appear to me to have been about
trendlines and regression curves, which are not covered by this issue.


**ALL MEMBERS ADDING COMMENTS PLEASE READ** Thank you

**It would be useful if future commenters could please reread the initial
Summary and Description at the top of the page, as you could be wasting your
time and that of the OOo team by putting your comments on the wrong issue. It
could also risk invalidating the original issue by diluting it with material
irrelvant to this issue, making it difficult to interpret, which would be a
great shame.**

You're right derekhtodd, this post is about error bars, but others issues refers
to this also for trend lines and their equations.

Regards,
ap

ap78, thanks for the information. The issue which appears appropriate for your
comments posted here, I think, is Issue 5289, and I see you have posted there.
If any non-related issues direct you to this one, than they are of course in
error. I'd be greatful if you could email me with the details of any examples
you have of non "define custom error bar range" related issues pointing to this
issue, so that things can be clarified. Many thanks,

Derek

I looked for them and found: 23909, 22740, 25277, 28458 and 40578. However,
you've already clarified that they are not a duplicate of 366 :-)

Thank you,
ap

Respectfully suggest increase in priority.  I am unable to use OpenOffice for my
undergraduate Modern Physics Lab course because of this lack of basic
functionality.  I had to install a copy of Windows on my laptop so that I could
use Microsoft Excel.  Graphs without error bars are useless to scientific users.

*** Issue 4826 has been marked as a duplicate of this issue. ***

Raised priority and changed issue type to feature.

*** Issue 58553 has been marked as a duplicate of this issue. ***

*** Issue 3380 has been marked as a duplicate of this issue. ***

As a biologist, I heartily agree with the comments above. Any scientist will
tell you that adding custom error bars is absolutely critical for presenting
data. Any  presentation of data without this "feature" would be immediately
rejected from publication and laughed out of a symposium. I can't use OOCALC's
chart feature for teaching purposes or for my own research for this reason. It
should be implemented immediately.          

Sincerely,
Chad Smith
University of Texas at Austin

Earth to OOCalc developers!

Please pick up a statistics book and look into the difference between regression
analysis and error bars - they are NOT THE SAME!

This has gone on long enough - 5 yrs and there still isn't someone at OO that
knows what the heck we're talking about.  Please pick up a book and read up on
what it is we're looking for or to make life easier then go find a scientist
that uses MS Excel and just ask them to show you what it is we're talking about.

And I thought MS was slow...

->enmane: What do you think will you achieve by your rant here? Of course we
know the difference between regression analysis and error bars. If would take a
look at the product you are talking about you would see that these are different
things and that they are, apart from some missing features (e.g. this one), the
same as in MS Excel.

I won't explain why we haven't implemented this feature as I think I have done
this often enough. We can start new features only when the basis for this has
been done, so you have to be patient. Ranting doesn't help. Helping in
development would. We are two people re-implementing the chart. So far in all
this years we just got one developer who just started some small implementations.

So may be I am allowed to rant, too. Why do you guys think when you get a
software for FREE and do NOTHING for it apart from ranting, that we should
listen to you. We will implement this feature, when we finished chart2 and have
the time for it. Period.

So, if anybody has some constructive advice, a specification for how to
implement this feature, you are welcome to help. If you think this issue is
important, you can vote for it, but as you can see a high vote count does not
mean that this is implemented immediately, because in the current chart
implementation it is impossible to implement this feature, so we all have to
wait until the re-implementation is finished before we can care about this issue.

And if you think we are stupid then either tell us what we misunderstood or shut
up, as this kind of rant doesn't help anybody. (It just takes time again to
write this comment, that could have been used for doing real work)

And what do you mean by "what we are looking for"? I think the summary is very
clear: this issue is about "Select 'Range' for error bars" and nothing else.

First of all, thank you bm and the OO team for your efforts in doing the
programming for the charting in OO. In regards to this issue, I think that there
is a high level of frustration among OO users due to the fact that this feature
is critical for serious use of OO in some fields, including scientific research.
This prevents me and many others to use and/or promote OO among colleages and
force us to go back to use MS Office, in spite of our preference for OO or
Staroffice. 
I understand that it is heavy work and I whish I could help, but I am not a
programmer. I concluded that the best way for someone like me to help is
topromote OO use, and that is why frustration is high, because I cannot until
issues like this are resolved.

Daniel

sorry if I've made you upset but there is no reason that my post on

Tue Feb 10 11:00:30 -0700 2004

should be in the same Issue 366.  That post was about regression analysis and
curve fitting and not error bars.  The fact that they were lumped together
implies that they are the same thing when they are, in fact, not.

Please see that post again and re-read it as I DID offer to help.  I am not a
great coder although I have coded in C, C++, Fortran, Pascal, and Basic but it
has been a while.  I do know regression analysis and statistics.  You let me
know what it is that you guys are lacking and if I can contribute then I will.

I agree 100% with Daniel - I've STOPPED promoting this package because it really
doesn't compare w/MS Office at all.  Sure it has a few features that are really
great and even better than MS Office but as a whole, it is still a decade
behind.  Now that 2007 is coming out, it'll fall even further behind.  

I still believe that the original implementation was vastly superior - 1 desktop
that handles everything - and Sun made a mistake in wasting all those resources
in splitting up the packages and beautification when it could have gone to added
real features that were lacking.  This clearly shows why Sun, as a company,
hasn't done well in decades.  Having 2 people working on this is ridiculous.

to clarify your post on 
"So may be I am allowed to rant, too. Why do you guys think when you get a
software for FREE and do NOTHING for it apart from ranting, that we should
listen to you. We will implement this feature, when we finished chart2 and have
the time for it. Period."

-->  Well, I know that I feel that my time is very valuable and I want to
promote OO but when I waste considerable time on features that don't exist then
it costs me money, if time=money.  The money that is spent isn't only the money
that is laid down for a package.  I feel lied to when the promotions come out
talking about how great OO is and I waste my time to download it and promote it
only to find out little significant (to me) progress has been made.  The fact
remains that people value their time and wasting it upsets me.  My reputation is
on the line when I promote this package only to find out that there are still
missing features.  I want to help and promote it but I won't do that anymore
because of the history of development and lack of features.  So YEAH, I feel
justified in bitching about it because OO has cost me considerable time,
resources, and has affected my reputation with regards to this package.  Not all
things are measured in greenbacks that have been pulled from one's wallet and
one could argue that the really important things aren't measured with greenbacks
at all.

BM, your efforts and the other person are greatly appreciated.  My issue isn't
with you but the fact that Sun isn't allocating their resources effectively on
the issues that are critical.

Sun should take this example into consideration,
Buy MS OFfice for $100 (very easy to find price)- all features included and provided
Person pays nothing for OO - cost $0 but spends at least 5 hours over the course
of 4 yrs finding ways to do things such as,
* creating good looking templates that already exist in MSO
* doing proper plotting
* productivity losses due to speed differences
* exporting filtering issues
* getting multimedia to play correctly in OOImpress (windows AND linux)
* figuring out how to get things done in OO vs MSO
At a cost of $20/hr (cheap, if you ask me) then the person would be better off
with MSO because they would have saved time and NOT been frustrated.  Now,
consider that a 5 hr loss over 4 yrs is very conservative, I know that I've
wasted that in a few months on OOImpress.  Consider that $20/hr is pretty cheap
her in the US and you'll see that OO has a very tough battle ahead of it.  If
there is any wonder why people aren't using it, just point the Sun executives to
this example.  
  I would be better off using MSOffice both for reasons of frustration and lost
time/money.  Based on this example, the only ones that would use OO are those
that just cannot afford to lay out MSO or those that want to promote it at the
expense of wasting their time (for whatever reason such as dislike of MSO and
wanting to see a true competitor); well, I know I fall into the latter camp
along with many of the posters here and we are walking away one after another
because eventually we realize that the shine on OO has lost its luster.
  So unless Sun makes some changes real quickly then the ones that are using it
but can afford it are going to walk away if they haven't already.  It just isn't
worth our time.  Heck, I know that I'd even offer $$ if the package was anywhere
near MSO but it isn't and I'm not about to give money to a company that doesn't
heed the community.  I've given considerable amount of money to opensource
efforts such as PCLINUXOS, Mandrake, and Mozilla (among some) - much more than
to MSO - but that is AFTER I got some value out of their product.  I usually do
it after 6 months of use or so and signs that they are developing in a direction
that I agree with.  I would imagine THAT is the reason that OO isn't seeing too
much money and Mozilla has so much money that they don't know what to do with it
all.

Well, just wasted another 30 minutes on this posting.  Heck, I bet I've wasted
my 5 hours on this 1 issue alone.  So now how is someone at Sun going to
convince me to continue using this product?  Because it's FREE?  IT ISN'T!  Not
so long as people value their time on this Earth.  

Don't take it personal bm, the issue is larger than that and rests with the
people leading Sun.  So why have I singled Sun out and not OO developers?  It's
because the majority of OO developers are actually Sun employees.

Please, everyone, any comments not of direct relevance to the techical aspects
of implementing this issue should be mailed directly, or taken to the mailing
lists, where they belong.

Issuezilla is not a discussion forum, and this is wasting everyones time and
benefiting nobody. If anyone feels that they absolutely must respond to this
(and I'd very much rather they didn't), please mail directly and not add more
unhelpful noise. And yes, I do realise that I've added to it, but hopefully to a
constructive end.

Thank you.

*** Issue 69423 has been marked as a duplicate of this issue. ***

*** Issue 69423 has been marked as a duplicate of this issue. ***

The barrage of comments over the past 3 or more years says it all.  If Calc is
to be suitable for serious scientific work, it must have error bar capabilities
at least equal to those of Excel, and must display error bars from an Excel
file. Is the continuing lack of this feature evidence of a disconnect between
the OOo team and scientists? 

It's more fundamental than that and has always been the problem when CS people
lose touch with what the clients are looking for.  

I've made myself available on numerous occassions as a numerical methods person
(not a coder) and I haven't heard a single thing from a single coder asking for
any input.  At this point, I don't care anymore.

I'm beginning to wonder that WHEN these features do appear whether or not
they'll actually be worth using based on a technical standpoint.  If the coders
aren't working with people that know how to do numerical methods then what are
the odds that it'll even come out correctly - this is a SERIOUS problem with MS
Excel and very few scientists use the results from Excel in technical journals
because of it's deficiencies.  A simple google search will show this.

->thfil, enmane: Reading all the comments in this issue it should be clear for
everyone what the situation is. Repeating the same stuff again and again doesn't
help anyone, it only increases the number of comments thus renders this issue
more and more useless. Please note again that this is not a discussion forum.

A MORE COMPREHENSIVE APPROACH
=============================

Many issues were discussed in this thread (beyond the original posting). Please
note however, that there is a more general approach to all the issues listed.

WE DO NOT NEED TO REINVENT THE WHEEL!

See also the wiki page on statistical methods
(http://wiki.services.openoffice.org/wiki/Statistical_Data_Analysis_Tool ,
especially the paragraph 6 - Task Breakdown - Third Party Library Integration).
and the issue http://www.openoffice.org/issues/show_bug.cgi?id=66589 ("R
Statistical Software", which aims at integrating R with OOo Calc).

Implementing all this issues directly into OOo Calc will be difficult.

DISADVANTAGES
-------------
 - many resources (coders, even financial constraints)
 - code becomes more and more complex, more difficult to understand by new
developers, more testing time needed
 - experienced coders work already on other open source projects; less
availability for a new project

ALTERNATIVE
-----------
 - a very good alternative is to do all these advanced things with dedicated
external software, when such free alternatives exist:
  -- I have read something about gnuplot on this thread.
  -- There are other softwares as well, e.g. R (with over 500 packages), octave
and many more, each suitable for some specific task.

ADVANTAGE
---------
 - one can easily implement many more features from the specific program with
minimal effort (once the external program is embedded within OOo)
 - dedicated programs offer more advanced solutions than anything that would be
implemented in OOo (or some competitive software)
 - less resources needed:
  -- programs are developed by their own groups,
  -- tested by appropriate folks and therefore
  -- NO major delays, less propensity for bugs
  -- new features are implemented more easily (we at OOo need only make a new
Menu/Gui and paste the correct syntax to the external program)
 - smaller code, smaller program
  -- should also run faster, not so many resources loaded
  -- only users that need that option will run it (see also the note below)


A special note:
 - because some licenses may not be compatible with OOo (although they are still
open source), I DO NOT MEAN including the code in OOo
 - what I mean is a general mechanism in place that allows OOo to communicate
(bidirectioanly) with the external software
 - instruct users what external programs (extensions) do exist and where to find
them (url)
 - allow users to easily acces the functionality of that programs through OOo
Menus (or functions):
  -- e.g. R is a state-of-the-art statistical software, but it is NOT
menu-driven and most users will find it difficult to use, so implement handy
menus in OOo that paste the correct syntax into R (R can be embedded in other
implications, see paragraph 7 - Links, on the wiki page).

There is a previous comment that fits very well to my post on external software
integration. (Sorry for the double post)

> I'm beginning to wonder that WHEN these features
> do appear whether or not they'll actually be worth using
> based on a technical standpoint.  If the coders aren't working
> with people that know how to do numerical methods then what are
> the odds that it'll even come out correctly ...

This is another strong reason why external software are a better alternative!

I can only partially agree with discoleo. Using external applications for
advanced statistical analysis is probably the best choice from a technical point
of view.

However there's no doubt that the absence of statistical tools in OpenOffice
represents a serious limitation for users and, as consequence, for the adoption
of OpenOffice in many work places such as labs, schools, universties.

I think more factors should be considered.
Using external applications requires the user to install and learn how to use
them, which is usually more difficult than working with a spreadsheet. For
example, doing a regression in Excel is just some click.

Also, another important factor is that many applications used to manage analysis
instruments in labs save data in a spreadsheet format (usually xls). So it's
immediate to manipulate data in Excel or Calc, while it would require
conversions to use external software.

**ALL MEMBERS ADDING COMMENTS PLEASE READ** Thank you

**It would be useful if future commenters could please reread the initial
Summary and Description at the top of the page, as you could be wasting your
time and that of the OOo team by putting your comments on the wrong issue. It
could also risk invalidating the original issue by diluting it with material
irrelvant to this issue, making it difficult to interpret, which would be a
great shame.**

*** Issue 71948 has been marked as a duplicate of this issue. ***

*** Issue 72322 has been marked as a duplicate of this issue. ***

I must agree that selecting a range for errorbars is the most important feature
I am missing in OOcalc. I do not want to repeat the arguments which are already
listed above.

Unfortunately I have no real programming skills to contribute to this issue. But
perhaps it is helpful to point out, that this "errorbar feature" is already
implemented in "gnumeric". The source code of gnumeric is free. I do not know if
it is possible, but it might be easier to transfer this code from gnumeric to OO
instead of reinventing the wheel.

Hi guys!
It's really strange that "custom error bars" feature still has not been 
implemented in OO. I am a system administrator in a big scientific institution 
and I can say that for now absence of this feature for us seems to be THE ONLY 
major factor hindering transfer from MS Office to OO. Using Calc without this 
feature in a scientific setting isn't worth using Calc at all, because, as 
opposed to common office tasks, research calculations almost always should 
result in "barred" charts with custom error bars (by the way, most scientific 
journals just will not accept an article if your chart doesn't have them). As 
it's been stated here already, scientific society is one of the most useful and 
perspective fields for spreading and implementation of OO. Well, I understand 
that the current "list of wishes" for Calc's improvements is quite big, but the 
discussed feature should be, of course, placed to the beginning of it.

3

gang, as much as I'm frustrated with the developers taking so long (6-7 yrs on
the weakest point of Calc) here's what we can do,
1) Use the point system and give this problem points so that your voices are
heard by the metrics they use.
2) Try out http://graphics.openoffice.org/chart/chart.html  - the coming new
charting utility which "promises" to help solve "some" of our problems. 
Download it and install it and report the bugs BEFORE they get into 2.3
otherwise we'll all be back here again.

They ARE making progress even though it might take longer than we'd like.

-> lunacarb: please avoid making "noise". Consider posting on the issue only if
you add real value to it. What coud be better, is ask how you could help to make
things faster. As enmane said, testing the new version is a way to help. 

-> enmane: thanks for your reply ;-)

-> pagalmes: May be I got it wrong, but it comes out that for the issue tracker 
the first post should be the only one, and anyone (excluding the initiator) 
shouldn't make "noise" and has to just vote if he/she agrees with it. What 
about discussing all pros and cons for the issue? Should it be held separately 
somewhere?

changed target to 2.x

This is the single issue which stops me from being happy with my linux system. 
I am a scientist and as many of my colleagues have already noted, Openoffice 
Calc is USELESS without this feature.

Please, its been more than 6 years since this was first reported.

Adding noise and pleas in this bug report only delays possible development, if
anything. Please keep silent (and just use the Vote for this issue -feature)
unless you have some real contribution to make, probably some code patch or
other thing that has taken you countless hours to develop and aids the real
development work in a significant way, while at the same time being aware all
the development that has been done in so far, ie. having read mailing lists for
some months etc. and being comfortable with the OOo code base.

Anyway, the content of this particular noise message is to re-iterate that
Chart2 has been under construction for a long time now, which may or may not
include all the Chart features people want, and may be possibly included in
OpenOffice.org 2.3, coming out in September. If you want to help in developing
to make sure "your" feature will be there, you join the mailing lists, get the
source code and develop. Adding comments in bug reports when the bug has been
already described does not help, quite the opposite.

Thanks. Hopefully this will be the end of noise. (unfortunately I've a feeling
it's not...)

mrv, 
many of us are not coders and can't help in this fashion.  I understand your
frustration but understand mine - I've offered to help a coder with the math
involved in regression analysis and haven't been taken up on my offer and that
says something about the "team."

To the rest of you,
  Download a copy of the currently-being-developed Chart 2 and try it out.  I've
already created 3 bug reports on it.  If you don't voice your opinions now then
it will most likely be another 6 yrs before you get your feature included.  In
summary, do the following:
1) VOTE - use your points to vote for the issues that are important to you
2) Stop beating this dead horse-they are working on Chart 2 so focus your
energies on fixing Chart 2
3) If you aren't happy with the current implementations and want to try your
luck on something being developed then work with Chart 2 which "should" have
some of the features your looking for and vote for the features that are missing!

Turn your frustrations into positive actions.  Play the game properly!

@enmane: I downloaded and tested the new chart, and I'm disappointed. But I also
gave up with calc anf I go on using Excel, which still is far superior in many
fields.

It's impossible after such a long time not to have basic (yes, they are basic
and not advanced as someone think) functionalies like non-linear regressions.
I'm not interested in seeing a plot of the regression, but in obtaining an
equation in some click without using a different application.

This and other features are postponed without a clear reason, and help from
users is not accepted if not in terms of an already ready solution.

It's really sad to say, but in this way OOo is going nowhere.

Regards

Developers, just to be clear, we users are well aware of the complexity of the
OOo codebase and the amount of work that it will take to implement this and
other essential features. We don't expect miracles.

We do expect an apology for not recognizing the seriousness of this problem, and
for not making it a higher priority earlier on. An acknowledgment that serious
mistakes have been made in the past would make a lot of scientific users much
more willing to support you in the future. Defensiveness does not breed the
things you need the most: evangelism and contributions.

Thank you for your work on this project. I use OOo for simple data analysis, and
supplement it with Matlab and statistical packages for more complex work, and
Grace for graphs. I look forward to being able to use Impress for scientific
presentations without having to import external images for graphs with error bars.

Please, stop flames !

@pagalmes

Sorry, I didn't mean to flame. I was just commenting and explaining my point of
view. 
I also recognize the good work done on other parts of OOo like writer, which is
superior to Word in my opinion, and thanks to Novell efforts now is more
compatible with the MS formats too.

Regards

setting target

Why not use RLPLOT as an add-on for scientific/statistical graphs?It would be
great if RLPLOT could be included in future versions of OOo.
There is a great scientific and academic community out there that is spending a
lot of money in MSExcel, why not helping them now:

Can anyone code an extension for using RLPLOT in combination with OOo 2.X? (Im
sure the rlplot people would be delighted to help with that)

http://rlplot.sourceforge.net/index.html
http://rlplot.sourceforge.net/Examples/index.html

Fixed in CWS chart20.

Is there a spec about how this was implemented?

Setting URL to Spec.

Thanks Björn !

->TK: Please verify in CWS chart20.

As this fix required some profound changes in the XML filter for charts with own
data, it is important to test load/save of charts with own data and clipboard
from Calc (charts taking data from sheets) to Impress thoroughly.

Created attachment 51811 [details]
Testcasespec for Y Error Bars

Seen ok in CWS chart20 -> verified

Can we get an idea of when this change will make it into the binaries? Is this
going to be a September '08, OpenOffice 3.0 thing, or will it be available
beforehand.

Thanks in advance, and thanks for your hard work. When this change goes through,
I will finally be rid of my need for Gnumeric to do data analysis. 

Bug 366. When I first reported it, a bit over 7 years ago, I thought I would 
soon switch to OO.org after this was fixed. Much has happened since then. I 
became so much involved in free software that I switched to openoffice.org 
anyway. Even although I did my graphs with other programs (gnumeric, for all 
has fallacies, had lately become the most easy tool for quick graphs though).

Actually, spreadsheets are not that suitable for publication quality graphs, 
but it is nice to have something for quick and easy data inspection or the 
weekly talk.

I wish to thank all the people that worked on this. I hope all my fellow 
scientist in this thread now follow up to their promises here and pollute all 
lab computers with OO.org 3.0. 

Cheers,

Danny

>I wish to thank all the people that worked on this. 
>I hope all my fellow scientist in this thread now 
>follow up to their promises here and pollute all 
>lab computers with OO.org 3.0. 

Nice thought, but not until ISSUE #42946 gets implemented:
http://www.openoffice.org/issues/show_bug.cgi?id=42946

Still a deal breaker for many scientists.

Sorry to post another comment to this bug, but since this bug is now "closed"
and it still has 131 votes, please consider moving your votes to ISSUE #42946
(error bars for x values) 

It is both an interoperability issue (as MSOffice has this functionality) and a
usability issue (as it limits the usefulness of OpenOffice for science students).

/soapbox

I have just read he comment from davidbl on RLPLOT and checked that out. It is
EXCELLENT! It's so much work already done!

Why not incorporate it into OO, either as a component, or as an add-on?

That would be marvellous!

Seen ok in current master -> closed