Where have markers to be drawn in custom-shapes?

Regina Henschel rb.henschel at t-online.de
Tue Sep 21 12:59:37 UTC 2021

Hi Miklos,

Miklos Vajna schrieb am 21.09.2021 um 08:35:
> Hi Regina,
> On Mon, Sep 13, 2021 at 07:35:10PM +0200, Regina Henschel <rb.henschel at t-online.de> wrote:
>> I cannot determine where to draw markers for the draw:enhanced-path of a
>> custom-shape. This question is not about path-objects and not about SVG, but
>> only about custom-shapes.
>> ODF has no rule for it yet.
> I would default to what LibreOffice does today, just to avoid
> compatibility problems. What do you think?

I think, that we should not use the current behavior of LibreOffice 
because it is inconsistent. I have already reported

I have created an OASIS issue
That has a file attached with problematic situation.

In my opinion there should be this behavior:
[In the following I mean "sub-path" in ODF sense, not in LO internal sense.]
(1) A sub-path which is closed by a command Z and not continued, should 
never have markers. Currently LO draws markers if there exists a 
sub-path after the closed sub-path which has e.g. a line. I mean command 
sequences like MLLZML.
(2) A sub-path which is closed and continued, should have marker start. 
This situation is currently wrong implemented. I mean command sequences 
like MLLZLL are wrongly drawn by LO. A command Z does not end a sub-path.
(3) A sub-path were accidentally "first point = last point" is currently 
drawn without markers (if not continued), although having "first point = 
last point" does not close the sub-path. The equivalence of "first point 
= last point" to "closed" is only an (problematic) implementation detail 
for path-objects, but an enhanced-path has a dedicated "close" command 
and the path might not end with the "close" command.
(4) Different sets of sub-paths should be treated independently. I mean 
sequences like MLLZNMLML.

In regard to "compatibility": MS Office behaves already different on 
ODF-files than LibreOffice, whereby I think, that MS Office is neither a 
role model as it has inconsistencies too. Different versions of LO would 
only be affected in non-primitive shapes. I think (but I have not 
examined all predefined shapes), the predefined shapes have no 
problematic path definitions. Closed sub-path and separate lines are in 
different sets, so that problem (1) does not occur. And situations (2) 
and (3) do not exist in the predefined shapes.

Kind regards,

More information about the LibreOffice mailing list