[Libreoffice-ux-advise] Chart: what to do when trendline (regression) equation is too long tdf#94004
Heiko Tietze
tietze.heiko at googlemail.com
Tue Jan 12 03:04:07 PST 2016
Replied on the bug tracker. Would be good to have a UX keyword to not only
get pinged by component=ux-advice.
2016-01-08 22:34 GMT+01:00 Laurent BP <jumbo4444 at yahoo.fr>:
> Hello,
>
> If your trendline equation is too long (for instance with a high degree
> polynomial trendline) then :
> - end of equation is not visible [1]
> - chart is squeezed when exiting chart [2]
> See bug 94004 [3]
> To fix this bug there could be (at least) two ways:
> (a) if General format is used for equation, number of decimal digits should
> be adapted to fit equation in chart width
> (b) we should alert user that equation is too long and let him make a
> decision: change number format, change font size, increase chart width
>
> I proposed a commit [4] for solution (a), but Moggi have some concerns
> about
> it and would prefer solution (b).
>
> moggi wrote
> > I'm not sure if that is really a good approach. In general I think
> instead
> > of trying to change the equation we should make it clear to the user that
> > we could not layout his chart correctly and that he should look into
> > fixing the layout. So a more make the user aware of the problems than
> > trying to work around problems.
> > I suggest that you contact the UX team and tell them about your idea and
> > my concerns.
> > [...]
> > If they have more questions they can contact me.
>
> IMHO, if General number format is used, it should automatically adapt its
> length to chart width like it does in a cell with column width. With
> General
> format, user does not bother about specific notation (normal or scientific)
> nor precision (as many as digits as it could). For instance, if intercept
> is
> forced to 1 (you can set its value through trendline dialog), there is no
> need for a large number of significant digits for intercept value; but
> other
> values of polynomial equation may display much more digits. Only General
> format can deal with different precision for each value of polynomial
> equation. So user could not fix the problem by means of a specific number
> format: all values of polynomial equation will have the same format.
>
> However, solution (a) does not fix the bug if user chose a specific number
> format that creates a too long equation. In this case, user is already
> aware
> of the problem when chart is opened (as it can be seen on screenshot [1]).
> And when chart is closed, I propose that trendline equation should be cut
> to
> chart width and ended with a red triangle (like when text length exceeds
> column width within a cell). So that chart is not shrink like in screenshot
> [2], and "make the user aware of the problems" with solution (b).
>
> So now we are expecting your advices about these two solutions.
>
> Many thanks in advance for your comments.
>
> Best regards,
>
> Laurent BP
>
> [1] https://bugs.documentfoundation.org/attachment.cgi?id=121807
> [2] https://bugs.documentfoundation.org/attachment.cgi?id=121808
> [3] https://bugs.documentfoundation.org/show_bug.cgi?id=94004
> [4] https://gerrit.libreoffice.org/18397/
>
>
>
> -----
> LibreOffice 5.0.4.2
> --
> View this message in context:
> http://nabble.documentfoundation.org/Chart-what-to-do-when-trendline-regression-equation-is-too-long-tdf-94004-tp4171208.html
> Sent from the UX-Advise mailing list archive at Nabble.com.
> _______________________________________________
> Libreoffice-ux-advise mailing list
> Libreoffice-ux-advise at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/libreoffice-ux-advise
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/libreoffice-ux-advise/attachments/20160112/c7b17994/attachment.html>
More information about the Libreoffice-ux-advise
mailing list