Danny Trebbien wrote: > >> Hi Danny. Please could you enhance your tests so that they can detect > >> this error (so that they will fail if this error is present). > > > > Sometimes I should use more words. > > > > What I meant to say is: I'm trying to devote as much of my time as > > possible to other things (WC-NG specifically) so if you could help me > > out as much as possible with reducing the review effort needed, I'd > > really appreciate it. I assume your test didn't pick up this error > > already, and I can't see at a glance whether you have extended the test > > so that it does. If you do that, then that'll save me (and anyone now > > and forever) from having to scrutinize the code to see whether it > > handles this properly. That precisely what a regression test is for. > > > > So if you already extended it, apologies as I haven't tried it or > > examined it, as I wanted to ask you first. > > > > Also please include the log message each time you attach a patch. > > Please could you re-post the patch and log message together even if no > > further changes are needed. > > > > Thanks, > > - Julian > > Hi Julian, > > I have already looked into adding a regression test for this, but it > is not possible even after applying the patch. The only way that the > problem would manifest is if one of the other svn_subst_translate* > functions that allow EOL_STR to be specified were also extended to > accept a TRANSLATED_LINE_ENDINGS parameter. For example, if > svn_subst_translate_cstring2() were extended in this manner, "\r" was > used as the EOL string, and the value to be translated included only > Windows-style line endings, then the problem would occur. With this > patch, only svn_subst_translate_string() is extended, and > svn_subst_translate_string() always uses "\n" for EOL_STR.
Oh, OK. Thanks for explaining. > Attached is the patch and log message. I'll get to this soon - hopefully today or tomorrow. - Julian
