On Aug 11, 2011, at 10:48, Bradley Giesbrecht wrote:
>>> +post-activate {
>>> + delete ${prefix}/etc/logrotate.d/.turd_${name} \
>>> + ${prefix}/var/run/logrotate/.turd_${name}
>>> +}
>>
>> It's a bit odd to be deleting the turds... does their presence really cause
>> a problem?
>
> I'm not sure about the var/run/logrotate turd, but the etc/logrotate file
> globs the contents of etc/logrotate.d. I think there was a problem with that.
> I'll remove it and test. This is to old for me to remember.
I'm guessing deleting the one in var/run/logrotate is not necessary, and that
deleting the one in etc/logrotate.d was because as you said the script finds
all files in that directory and tries to run them. If that's so, then there
might be other ways of fixing it; for example, the script could only try to run
files that have the execute bit set.
>> But moreover, you're offering the user a choice between gzip and bzip2. (And
>> not a choice between bzip2 on or off.) In other words, if considered in the
>> context of a GUI preferences dialog, this choice might be represented as two
>> radio buttons; it would not be represented as a checkbox. Standalone
>> variants usually represent checkbox-like on-or-off choices. To implement
>> radio-button-like choices in variants, I find it makes more sense to make
>> two variants that are conflicting with one another. To do this, remove
>> port:gzip from the port's global depends_run, and then:
>>
>>
>> variant bzip2 conflicts gzip description {Use bzip2 compression by default} {
>> build.env-append COMPRESS_COMMAND="${prefix}/bin/bzip2" \
>> COMPRESS_EXT=".bz2" \
>> UNCOMPRESS_COMMAND="${prefix}/bin/bunzip2"
>> depends_run-append port:bzip2
>> }
>>
>> variant gzip conflicts bzip2 description {Use gzip compression by default} {
>> depends_run-append port:gzip
>> }
>>
>> if {![variant_isset gzip]} {
>> default_variants +bzip2
>> variant_set bzip2
>> }
>>
>>
>> The variant_set line ensures that even if the user requests to install the
>> port with neither variant (-gzip -bzip2), the bzip2 variant will still be
>> used. (It doesn't make sense to have "all radio buttons off".)
>
> Ah, right. So using variant_set prevents -bzip2, right?
I would say it overwrites / supersedes it. The order seems to be first
default_variants makes a suggestion. Then the user's choices on the command
line and in variants.conf are evaluated. Finally variant_set can trump any of
those.
_______________________________________________
macports-dev mailing list
macports-dev@lists.macosforge.org
http://lists.macosforge.org/mailman/listinfo.cgi/macports-dev
