conflicting changesets for __go_draw_axes__

[Ben Abbott]

On Jun 18, 2009, at 10:29 PM, Ethan Merritt wrote:

> On Thursday 18 June 2009, Ben Abbott wrote:
>> I just noticed an unexpected conflict between two changesets.
>>
>> The one below changed the format for tick labels for log scale to be
>> "10^{%T}"
>>
>> 	http://hg.savannah.gnu.org/hgweb/octave/rev/f2152fad3563
>>
>> This more recent one dropped the "*" from the x11 fontspec (as  
>> gnuplot
>> passed this to x11 which apparently results in a search through its
>> fontpath for each text object with that fontspec. Hence some
>> complaints that plotting was slowed down.
>>
>> 	http://hg.savannah.gnu.org/hgweb/octave/rev/a00e219c402d
>>
>> I see no reason why changing the gnuplot fontspec from "*,12" to ", 
>> 12"
>> would impact text objects with a format specification of "10^{%T}",
>> but when both of these changes are present, the log scale axes label
>> are always "10" with no exponent.
>>
>> Using the current developers sources, this problem can be seen by ...
>>
>> 	clf
>> 	semilogy (1:100)
>>
>> Petr / Ethan, I've cc'd you in the hope one of you can give me some
>> advice as to why this is happening and how to fix it (I see it in  
>> both
>> a recent copy of gnuplot's developers sources as well as in 4.2.2).
>>
>> Ben
>
> I can't reproduce it here from inside gnuplot.
> set log y
> set format y "10^{%T}"
> set term x11 enhanced
> plot x^2
>
> show the expected superscripts.
>
> set term x11 font "*"
> or
> set term x11 font ",12"
>
> also works as expected

I should have duplicated the effect in a short gnuplot script first.

The one below does not display the exponents.

set terminal x11 enhanced
reset;
set format x "%g";
set xtics font ",14";
set log y
set format y "10^{%T}";
set ytics font ",14";
set xrange [0:100]
plot x

However, this one does

set terminal x11 enhanced
reset;
set format x "%g";
set xtics font ",12";
set log y
set format y "10^{%T}";
set ytics font ",12";
set xrange [0:100]
plot x

I assume that gnuplot is trying to render the exponents in both cases,  
but when the point size does not exist, nothing is rendered?

Ben