Could you please add a similar Processing Requirement for <ph>? It should explain why <ph> MUST NOT be used for well-formed spanning code (opening and closing tag are represented by separate <ph>).
For example if it's used to represent <bold> from HTML, agent cannot validate (without additional editing hints):
- correct order, which can produce </b>…<b>
- removal of either opening or closing tag
- tag nesting
- leaving tag pair empty
In case Extractor decides to drop opening tag at the beginning of the segment as "redundant", such as
<source>bold<ph id="1" disp="/bold"> text</source>
a correct formatting cannot be guaranteed if word order is different in target language
<target>text tučným<ph id="1" disp="/bold"></target>
Similarly, Processing Requirements should describe that <ph> should not be used for locking of inline content otherwise represented by well formed spanning code.
E.g.
<span translate="no>some text</span>
as
<data id="d1"><span translate="no>some text</span></data>
…
<source>following should not be localized: <ph id="1" dataRef="d1" /></source>
<mrk translate="no> should be suggested as preferred solution.
Thanks,