On Tue, Jan 24, 2012 at 4:41 PM, Richard Heck <rgh...@comcast.net> wrote:
> On 01/24/2012 02:13 AM, Abdelrazak Younes wrote: > > On Sat, Jan 21, 2012 at 10:02 PM, Richard Heck <rgh...@comcast.net<mailto: >> rgh...@comcast.net>> wrote: >> >> >> Consider: >> >> void Text::setFont(Cursor & cur, Font const & font) >> { >> LASSERT(this == cur.text(), /**/); >> // Set the current_font >> // Determine basis font >> FontInfo layoutfont; >> >> Wouldn't it be better to return here if the assertion fails? >> Continuing seems like we're inviting disaster, whereas returning >> just aborts what we're doing. >> >> >> Yes >> >> >> I've seen this same pattern a lot of places. I wouldn't propose to >> change all of them at once, but if this seems right we could try >> to change them as we see them. >> >> >> IIRC, Andre introduced this fallback code in LASSERT. The goal is >> eventually to have a fallback code everywhere, but not necessarily only >> "return". >> >> Right, so the default, when he did the change, was just to proceed. I > assume that is what we do if there's only a comment? > Exactly. Cheers, Abdel. > > Richard > >