0.10.12 — a guard that looks for characters

An hour after the previous release a second screenshot arrived. Beside the author’s rating stood the label “wstępna”, and under the comment form the whole consent notice about email contact — in Polish, on the English site.

The reason for missing them is instructive and worth recording. I searched with an expression that matches text inside quotation marks. A string standing directly in the view syntax has no quotation marks — it is the content of an element, between tags. So I found everything that was a literal, and nothing that was text.

Ten strings were added: the provisional-rating label, three accessibility labels, two states of the notifications button, the privacy note, the consent notice on the comment form, and the period abbreviations in the event log.

And a script was added that searches **by characters, not by quotation marks**: every Polish letter outside a comment is suspect. Comments stay in Polish — that is the documentation language of this project, so the script strips them before checking. It proved useful immediately, catching five strings in the instance’s own routes that nobody had touched during translation.