Re: ensure that generated files are read-only

[Simon Josefsson]

Bruno Haible <address@hidden> writes:

>> Note that this does affect modules/* files owned by others.
>> If anyone objects, I'll quickly revert the objectionable change.
>
> Please revert. It is not acceptable for me to have read-only files in a
> gettext or libiconv distribution.

I agree with Bruno here; I dislike read-only files as well.

In some of the cases committed here, the generated file that is
chmod'ed read-only are one-liners that are not unlikely to be modified
manually by users on, for example, MinGW, to work around problems in
the build environment or whatnot.  Having a one-liner file be
read-only serves no point.

Bruce Korb <address@hidden> writes:

> Jim Meyering wrote:
>> Considering the numerous uses of sed and echo (in place of cp)
>> I preferred to be consistent.  But I agree that it'd be better
>> not to pollute the build output with output that is 99% irrelevant.
>> So, how about prefixing each new line with "@", so make doesn't print them.
>> E.g.,
>>
>>   sys/select.h:
>>           test -d sys || mkdir sys
>>           @rm -f address@hidden $@
>>           echo '#include <sys/socket.h>' >address@hidden
>>           @chmod a-w address@hidden
>>           mv address@hidden $@
>
> My very strong preference is to not hide commands.  Disk space
> (log files) is cheap.  I can filter out chaff much more easily
> than I can reconstruct commands that are not in the log.
> No hidden commands without a good reason for hiding the command.
> Please.  :)  Cheers - Bruce

I happen to agree here too.  When reading build logs from remote
machines, every little piece of information helps.

/Simon