сб, 24 янв. 2026 г. в 21:59, William Lallemand <[email protected]>:
> On Sat, Jan 24, 2026 at 08:20:11PM +0100, Илья Шипицин wrote: > > it is not a problem. > > you can specify either > > > > * text=auto > > > > pr (for some files) > > > > *.c text=crlf > > > > and line ending will be handled as you;ve specified in .gitattributes > > no need to blame developers on windows for not configuring line endings. > > > > I'm not blaming anyone, but that's not the point, there are cases where you > need both LF and CRLF lines, and if you're not seeing them you won't get > what's > going on in the file. I don't think trying to automate the conversion is a > good > idea, maybe we could put the .c and .h in lf mode. > *.c and *.h seem to be good with "text=auto" > > > I sent a patch previously which accidentally changed line endings in > *.http > > files. I agree that per RFC line ending is supposed to be CRLF, but Willy > > told "You don't see them, but it has added CRs at the end of each line > above > > for no reason.", I do not argue with Willy :) > > I agree with Willy on this patch review, and I think you are missing the > problem : > > % cat -e examples/errorfiles/400.http > HTTP/1.0 400 Bad request^M$ > Cache-Control: no-cache^M$ > Connection: close^M$ > Content-Type: text/html^M$ > ^M$ > <html><body><h1>400 Bad request</h1>$ > Your browser sent an invalid request.$ > </body></html>$ > $ > > Your patch converted the LF from the body to CRLF. These files already had > CRLF > lines for headers, and LF for the body. So if your editor does not show > them, > you won't be able to edit the file correctly. An automatic conversion can't > help you there since it's a mix of both. > mix can be handled with "*.http binary" maybe, but let us postpone this > > > we can set crlf if it appropriate (to prevent accident line ending > convertion) > > I don't think it is useful unfortunately. > mix of LF and CRLF is not handled in my opinion (in automatic manner) > > Regards, > > -- > William Lallemand >
