[test, others..] EOLs and such.
I was looking at Boost.Test.. did an update.. and noticed the EOLs got messed up recently. So I ran the inspection tools and it found a variety of errors: http://www.boost.org/regression-logs/inspection_report.html -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq
"Rene Rivera" <grafik.list@redshift-software.com> wrote in message news:41ED3B7D.4080006@redshift-software.com...
I was looking at Boost.Test.. did an update.. and noticed the EOLs got messed up recently. So I ran the inspection tools and it found a variety of errors:
libs/test/example/.cvsignore: nonportable path
So?
libs/test/example/unit_test_example3.cpp: invalid (cr only) line-ending
'cr only' is UNIX format, is it? Gennadiy
Gennadiy Rozental wrote:
"Rene Rivera" <grafik.list@redshift-software.com> wrote in message news:41ED3B7D.4080006@redshift-software.com...
I was looking at Boost.Test.. did an update.. and noticed the EOLs got messed up recently. So I ran the inspection tools and it found a variety of errors:
libs/test/example/.cvsignore: nonportable path
So?
Good question ;-)
libs/test/example/unit_test_example3.cpp: invalid (cr only) line-ending
'cr only' is UNIX format, is it?
UNIX format is LF only. Mac, Amiga, some others are CR only. DOS is CR-LF. In this case since I ran the tests on a Windows machine it's reacting to EOLs that CVS has translated into: CR-CR-LF. So it sees the first CR as invalid (cr only). -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq
libs/test/example/unit_test_example3.cpp: invalid (cr only) line-ending
'cr only' is UNIX format, is it?
UNIX format is LF only. Mac, Amiga, some others are CR only. DOS is CR-LF.
In this case since I ran the tests on a Windows machine it's reacting to EOLs that CVS has translated into: CR-CR-LF. So it sees the first CR as invalid (cr only).
Are you sure it's this way in cvs? Did you try to use cvs on Windows with UNIX mode? It may be the result of update in Windows. I couldn't imagine how I could've put CR-CR-LF into files. Gennadiy
Gennadiy Rozental wrote:
libs/test/example/unit_test_example3.cpp: invalid (cr only) line-ending
'cr only' is UNIX format, is it?
UNIX format is LF only. Mac, Amiga, some others are CR only. DOS is CR-LF.
In this case since I ran the tests on a Windows machine it's reacting to EOLs that CVS has translated into: CR-CR-LF. So it sees the first CR as invalid (cr only).
Are you sure it's this way in cvs?
All I'm sure of is that it's not LF in the CVS server as it's supposed to be.
Did you try to use cvs on Windows with UNIX mode?
Absolutely not. That would ruin my checkout. And make editing with Windows programs harmful.
It may be the result of update in Windows.
What do you mean by "update on Windows"? I got the the errors because my WinCVS client did a LF to CR-LF translation as it's designed to do. And very likely the CVS server has files that are CR-LF, which translates to CR-CR-LF. This can only happen if someone puts in files with a CVS client that thinks it's putting in one type of file, but the files are really of some other type. Previously this tends to happen when someone edits files that are mounted on a shared drive from another OS. For example mounting a DOS drive in Linux.
I couldn't imagine how I could've put CR-CR-LF into files.
It's not that you put in CR-CR-LF, it's that you inadvertently lied to your CVS client about what kind of file you put in. Suspect would be if you recently used a different editor than you usually use. Or if you are mounting other foreign OS drives, either directly or networked. ...Of course none of this would be a problem if CVS was actually intelligent about EOLs :-( -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq
It's not that you put in CR-CR-LF, it's that you inadvertently lied to your CVS client about what kind of file you put in. Suspect would be if you recently used a different editor than you usually use. Or if you are mounting other foreign OS drives, either directly or networked.
Ok. I will look into it.
...Of course none of this would be a problem if CVS was actually intelligent about EOLs :-(
Actually I believe we could remedy this with server-side pre-commit scripts. Anybody considered this? Gennadiy
participants (2)
-
Gennadiy Rozental -
Rene Rivera