Translators can not use literal numerals in translations instead of variables as placeholders for numbers.
The reason behind this bug is to really fix an old issue we have with translations in Romanian language. This is a continuation of this bug in bugzilla: https://bugzilla.gnome.org/show_bug.cgi?id=666434
As translators we need to differentiate between numbers in feminine form and masculine form. This is essential within screen readers like orca. The issue we encounter with Romanian language is that the singular form for the number "one" (often represented by the digit "1") can be read out aloud in two, four different ways: "un" (short form) or "unu" (long form) for masculine and "o" (short form) or "una" (long form) for feminine.
The issue arises within strings like the following: Extract from orca/po/ro.po:
#. Translators: This message is presented when the user is in a list of
#. shortcuts associated with Orca commands which are not specific to the
#. current application. It appears as the title of the dialog containing
#. the list.
#: ../src/orca/messages.py:2538
#, python-format
msgid "%d Screen reader default shortcut found."
msgid_plural "%d Screen reader default shortcuts found."
msgstr[0] ""
"%d scurtătură de tastatură implicită găsită pentru cititorul de ecran."
As long as the variable "%d" is required to be present in the translation string, orca (and some human readers) would most likely fail to to figure out which of the 4 forms for "1" to say. In this case it should be "o scurtătură de ...", but since it would be represented as "1 scurtătură de ..." the screen reader (and some people) would start reading "unu scurtătură de ...".
I believe this issue should be raised at a more global cross-project issue within the Gnome internationalization efforts. Thank you, Daniel Șerbănescu, Coordinator for Romanian Team