Alignment of objects changes by 1 px?

[LaRetta]

Brand new file created for this demo.  Outer rectangle is set to 169pt and inner is set to 36pt.  When I select them both and align both horizontally and vertically, it seems fine.  However, if I click again, it moves.  Why?

Would not FM select the best alignment the first time and thus repeat the same 'test' the second time?  Wouldn't its first choice be its second choice when selected again?  It does not matter if I deselect it in between tests ... it still jumps 1pt.  Attached is video showing the behavior and the file itself.

Yes, it seems a little thing ... but what if I am aligning many objects this way.  According to this demo, they may not line up after I am done.  The center rectangle moves from 197 to 198 (top) when selecting horizontal align.  And the center rectangle moves from 256 to 257 (left) when selecting vertical align.

Why won't it set it 256.5 x 197.5 the first time and leave it there?  How can it provide two DIFFERENT 'best center' choices?

Or is it only on 10.11.6 (El Capitan)?  Tried in both 15.0.1 and 14.0.6  I would like others' opinion with different OS/versions before I report it to FMI if any of you would be so kind.  :-)  I also just want to report it so others are aware of the oddity.  

MisAlign.fmp12

MisAlign.mp4

[webko]

It happens for me on Yosemite in FMP15 - and clicking Align... again puts it back - so my guess is that it assumes you want the other (equally) valid Align if you use the function, and then the other (original) when you use the command again....

[LaRetta]

17 hours ago, webko said:

...it assumes you want the other (equally) valid Align if you use the function, and then the other (original) when you use the command again....

Hi Webko!  Thank you for testing - I appreciate it!   But neither IS an 'equally valid Align'.   The valid align would move the inner rectangle to left 256.5 by top 197.5.and it should be repeatable no matter how many times you click it.  

It isn't only when selecting again ... if I sign off and back on, it selects the 'other' position, thinking, "oh, you don't like the 'center' you had last time?  Try this one" and it moves it again. We can never verify if something is aligned by repeating the action because when we do, we move it, which can now make it misalign with other objects we've positioned. 

Even if FM decided that it did not want to place an object at the .5 location, its rules should decide whether to always round up or down - not randomly one or the other.  One can clearly see the difference if you open this file in 13.  The result is the inner rectangle properly moves once to a .5 position and doesn't move again because it lands at the correct coordinates the first time.

This 'lazy' repositioning to whole pt (in 14 and 15) is obvious as well.  Attached is screen shot where the bars adjusted vertically to (lazy) whole pt instead of (proper) fraction, making it obvious that they are not positioned properly.  Can you spot the single extra pt between one of them?  If FM used fractions here, there still might be a slim difference but, as in older versions, it would not be obvious.

Edited August 25, 2016 by LaRetta

[Lee Smith]

Hi LaRetta,

Sorry I meant to write back yesterday and let you know that I can confirm what you are seeing.

Wow, and I thought it was me when it happened in my files.

Lee

[Ted S]

Happens with FMP15A on Windows 10 (1607) as well.

[LaRetta]

Thank you all for the confirmation.  I plan to submit this to FMI.