[Gambas-user] COPY does not always copy chgrp and chown

[Benoît Minisini]

> Benoît Minisini schrieb:
> >> Just found this (Gambas 2 project) when copying several students from
> >> one class into another. For each student, there are about 6 files to be
> >> copied from one directory into another. The first one will always get
> >> standard chgrp and chown of the application's user. The other files
> >> SOMETIMES are copied with correct group and ownership, MOSTLY not.
> >>
> >> All files have special group and ownership, controlling if teachers or
> >> only the office can access the data.
> >>
> >> Now, this is the function that does the actual copy:
> >>
> >>
> >>
> >> 'kopiert alle Dateien eines Namens in eine andere Kartei
> >> '- ziel ist der Zielpfad
> >> '- wenn eine gleichnamige Datei schon vorhanden ist, kommt "-1" zurück
> >> '- wenn es geklappt hat, kommt "0" zurück
> >>
> >> 'wird z. B. von dlgNamenKopieren benutzt
> >>
> >> PUBLIC FUNCTION NamenKopieren(schlyssel AS String, ziel AS String) AS
> >> Integer
> >> DIM original AS String
> >> DIM odatei AS String
> >> DIM opfad AS String
> >> DIM datei AS String
> >>
> >>    original = getFilename(schlyssel)
> >>
> >>    odatei = file.BaseName(original)
> >>    opfad = file.Dir(original)
> >>
> >>    IF Exist(ziel &/ odatei & ".felder") THEN
> >>      RETURN -1
> >>    END IF
> >>
> >>    FOR EACH datei IN Dir(opfad, odatei & ".*")
> >>      COPY opfad &/ datei TO ziel &/ datei
> >>    NEXT
> >>
> >>    RETURN 0
> >>
> >> END
> >>
> >>
> >> Do you see anything wrong here? As it sometimes works correctly I would
> >> actually think that it's a bug in copy.
> >>
> >> The filesystem is ext3.
> >>
> >> Regards
> >>
> >> Rolf
> >
> > COPY actually does the same thing as opening the input file, opening the
> > output file, reading the data, writing the data, closing the input gile,
> > closing the output file.
> >
> > So the output file will have the same authorizations as the input file,
> > except that they are filtered by the "umask" (man umask) and that the
> > file will be owned by the process who wrote the file.
> >
> > COPY is there just for simple cases.
> >
> > I suggest you use the "cp" command, using the "--preserve" option.
> >
> > Regards,
>
> Ahh - ok. Wouldn't it be possible to pimp COPY giving it a special
> "preserve" option which would take care about this?
>

Yes but no. Why trying to rewrite "cp"? I even don't know why I wrote COPY. 
:-)

I don't want the interpreter to be too big. So maybe a component dedicated to 
file operations?

-- 
Benoît