View previous topic :: View next topic |
Author |
Message |
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Wed Aug 26, 2020 12:20 am Post subject: Example Plot - Incorrect Plot ? |
|
|
EX 2 here:-
https://silverfrost.com/ftn95-help/clearwinp/simplepl/native.aspx
Produces the following graph (on the right hand side below)
and not the one in the documentation (which is on the left above)
Why ?
Note also the rather arbitrary axes spacing on the Y axis.
And the non-centred x-axis labels, and only every other one. Why ? _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... " |
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Sat Sep 05, 2020 8:37 pm Post subject: |
|
|
I'm surprised silverfrost haven't been able to reply to this question. _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... " |
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Tue Sep 08, 2020 5:40 pm Post subject: |
|
|
no one at SF able to shed any light on this one then ? _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... " |
|
Back to top |
|
|
Kenneth_Smith
Joined: 18 May 2012 Posts: 697 Location: Hamilton, Lanarkshire, Scotland.
|
Posted: Thu Sep 10, 2020 1:41 pm Post subject: |
|
|
I think this is a consequence of the minor changes that had been made to the native %pl. If you use the latest dlls you get a slightly different plot to the two examples in John's post.
https://www.dropbox.com/s/wdfl0x8pmc3di4c/pl_plot.jpg?dl=0 |
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Fri Sep 11, 2020 6:05 pm Post subject: |
|
|
looks to me like your plot is the same as mine on the rhs of top post here Ken.
I know something must have changed which is why I posted.
It's a bit disconcerting that this can happen.
The code isn't 'robust' imo in this respect.
I went to this example with a view to running it through it's pacesafter recent updates to various aspects of %PL but have been bodychecked at the first hurdle
I might carry on nd see what other things might happen, but I guess I might just provoke further oddities to appear unless SF manage to pin down what's causing this instability.
I've understood Paul's previous comments about ways to fudge the program by fixing various parameters but it's not always easy if you've got lots of plots to do if the program is 'sensitive'. _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... " |
|
Back to top |
|
|
Kenneth_Smith
Joined: 18 May 2012 Posts: 697 Location: Hamilton, Lanarkshire, Scotland.
|
Posted: Fri Sep 11, 2020 6:56 pm Post subject: |
|
|
John,
Look at the numbers on the x axis in your plot and compare with the one i produced.
Ken |
|
Back to top |
|
|
LitusSaxonicum
Joined: 23 Aug 2005 Posts: 2390 Location: Yateley, Hants, UK
|
Posted: Fri Sep 11, 2020 11:13 pm Post subject: |
|
|
Sometimes the difference in appearance of a window is the result of a Microsoft change, as exemplified by what happens between Windows versions, and now that Microsoft intend to do these changes via the updates, presumably they will happen whether you like it or not. I suppose that refinements in FTN95 will also make changes, just as if you changed your Fortran code.
I find that the differences show up most obviously when you have screen-grabbed something to go in a Help file, or in a Powerpoint file giving instructions on how to use a feature. Such an effect is pretty obvious in the FTN95.CHM or in the PDFs of the original printed documentation. They are frustrating, but are artefacts of �progress� and just have to be lived with.
I find it particularly frustrating that the appearance of all the %ib options are �out of date�, and the %tb (which %ib was supposed to replace, according to the documentation) is more easily adapted to the latest windows visual conventions, which for me involves the redesigning of hundreds of bitmaps. And that�s for a Microsoft change.
Eddie |
|
Back to top |
|
|
DanRRight
Joined: 10 Mar 2008 Posts: 2826 Location: South Pole, Antarctica
|
Posted: Sat Sep 12, 2020 8:18 pm Post subject: |
|
|
John, you are a perfectionist in this respect. You notice things i often do not pay attention. If Y axis name would be placed in the axis center (this is older defect which still exists in all versions of new native PL) i'd from first glance not noticed any defects. Catching such problems is the software QA (quality assurance) job and is currently popular by the way
I remember how you have ones noticed not centered by the tiny tiny margin the X axis numbers. Later on i also noticed that in some situations. Here 200, 400, 600 are shifted a bit but this is not noticeable. When you have just one digit numbers though this tiny shift is clearly visible. The same with Y axis numbers where they are shifted a bit down, specifically visible with negative numbers -0.3, -0.6, -0.9 because minus sign is not aligned with the tic mark. This is clearly systematic shift and hope it could be fixed quickly.
I do not find missing numbers on X axis is a problem, quite opposite but if being picky even more the appropriate ticks have to be smaller (they are called minor tick marks) . And distance of Y axis numbers and tics a bit larger.
Keep doing such important job, John. Silverfrost has to encourage it more, this improves software quality while people are doing this like the devoted fans, and for free.
Anyway, one small my own wish here too: do all see that the step on Y axis is 0.3 ?
0...0.3...0.6...0.9...1.2...
This is i'd consider as a real defect of the algorithm if there is no manual setup for that (before there was no such option, or it did not work reliably when tried, now i need to check again). The step=0.3 will not allow to have on axis plotted value of 1 which is most important in all used graphs in science and engineering. With huge probability people will use normalized values to plot, and there 1 on axis is literally a must. So the better automatic delta placement would be 0.1, 0.25, or 0.5 not the 0.3. If this would require too much time to fix (i know that could be painful, i also had 0.3 on my own graphics routines i wrote decades ago and hated that for many years), then there have to be the manual settings for step |
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Thu Sep 17, 2020 6:48 pm Post subject: |
|
|
Ken, thanks for the feedback
Your plot is a third version !
Look carefully and you'll see the x-axis values are not centred on the tics !
what's that all about !!
Probably one of the craziest things about the auto-ploting is the scales it often chooses !!! 0.3 on the Y axis ! not the best
In my running of the sample code it puts markers every 100 on the x-axis but only labels every 200 !!!!!!! what's all that abut !!! but at least they're centred ! _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... " |
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Sun Sep 20, 2020 6:34 pm Post subject: |
|
|
Dan wrote:
Quote: | John, you are a perfectionist in this respect. |
yes Dan, ever since things which should have been corrected have been left to stagnate !
This observation only came about because I was going to run some variations on this test problem, which I'm loathe to do now.
Paul has been clear in the past that he hasn't any intention to iron out the remainng 'bugs' - that's his choice and prerogative - but it just means a lot more work for the programmer to get their philosophy right.
As I see it, from my observations, you can have a program working fine, release it, and then a short time, or a long time, later all hell might break loose when the dataset changes.
I guess I just can't understand why these things are apparently so difficult that they can't be fixed.
In my opinion it should have been priority on the shopping list when the question was first raised in SF Towers back in March this year when Boris went all 'lockdown' ...
'What we gonna' do now ?'
My personal opinion is that it can only end in tears once the opposition get wind, if they haven't already. _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... "
Last edited by John-Silver on Mon Oct 12, 2020 4:44 pm; edited 1 time in total |
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Sun Sep 20, 2020 6:41 pm Post subject: |
|
|
I might go off and try out some of those log examples we were tinkering with in the early days "-' years back ..... but more and more I find myself asking the same question ... is it worth my effort ? _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... " |
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Fri Sep 25, 2020 1:51 pm Post subject: |
|
|
Ken, when you wrote:
Quote: | I think this is a consequence of the minor changes that had been made to the native %pl. If you use the latest dlls you get a slightly different plot to the two examples in John's post |
did you mean the DLLS here ? :
http://forums.silverfrost.com/viewtopic.php?t=4245 _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... " |
|
Back to top |
|
|
Kenneth_Smith
Joined: 18 May 2012 Posts: 697 Location: Hamilton, Lanarkshire, Scotland.
|
Posted: Tue Sep 29, 2020 4:39 pm Post subject: |
|
|
Yes, those are the dlls I am referring to. |
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Mon Oct 12, 2020 4:52 pm Post subject: |
|
|
On the 17th Sept I wrote:
Quote: | In my running of the sample code it puts markers every 100 on the x-axis but only labels every 200 !!!!!!! what's all that about !!! but at least they're centred ! |
I find that kind of change somewhat troubling !
I also note that when I wrote:
Quote: | ... but at least they're centred ! |
I made a faut-pas because:
a) on the original documented plot the very first label isn't centred
b) on 'my' current plot (top of this post, on rhs), the labels , besides being only every other one, are NOT centred - neither are Ken's 'latest' ones either, whereas the original plot they were perfectly centred !!!
- what the heck is that all about !!!???
It just makes me wonder if I risk having different formatting every time I run with a different dataset.
It surely can't be that difficult/time consuming to iron out these niggly bugs ?
My additional observations might give Paul & Co some inspiration (and keep them out of the pub ! :O� )as it's likely that the 'lockdown' will soon be re-introduced as I type this message! _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... " |
|
Back to top |
|
|
John-Silver
Joined: 30 Jul 2013 Posts: 1520 Location: Aerospace Valley
|
Posted: Thu Oct 29, 2020 7:27 pm Post subject: |
|
|
can any of the SF 'godfathers 'shed any light on this please? Ta.
And shouldn't the documentation be changed , _________________ ''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... " |
|
Back to top |
|
|
|