Re: [RFC] Improved Linux process title support in the CLI SAPI

Hi, could you rename these functions to: cli_process_set_title cli_process_get_title http://www.php.net/manual-lookup.php?pattern=set_&scope=quickref gives me a lot of hits like: xml_parser_set_options stream_set_read_buffer pg_set_info ps_set_border_color etc. its also a lot more readable as it is: object_action_subject. Am 2013-02-07 20:59, schrieb Keyur Govande:

Addressing a bunch of comments at once: 1) I've removed is_available and renamed the set/get methods to: cli_process_title_set/get. I've also removed the test dependencies on pcntl and posix. The patch is updated: https://gist.github.com/keyurdg/4728770 2) Updated the RFC's introduction section with more concrete details about why this is necessary In terms of doing this as an extension: the RFC addresses why this isn't possible. For folks who will not be using this feature, at worst they'll loose a few KB of memory: the amount needed to store the original argv and the original environ. On Thu, Feb 7, 2013 at 1:25 PM, Nikita Popov <nikita.ppv@gmail.com <mailto:nikita.ppv@gmail.com>> wrote:

    On Thu, Feb 7, 2013 at 2:39 PM, Christoph Rosse
    <crosse@2bepublished.at <mailto:crosse@2bepublished.at>>wrote:

why wouldn't this go into core? setting the name of the current php-process is definitely something everyone that develops

    php-cli scripts

could use.

    I use a lot of php-cli scripts and I've never seen the need. Without
    having
    hard data to back this up, I am pretty sure that this applies to
    nearly all
    php-cli scripts.

We should not base the decision of putting something into the core on assumptions on how many people are going to use the feature.

    Obviously we should. Whether people will use it is pretty much the most
    important aspect for deciding whether or not something should be added.
    Even a trivial addition is a loose for the project if nobody is going to
    use it. And this is no trivial addition. This seems to be quite a bit
    system dependent and uses some odd methods like overwriting argv memory.
    And on that note, it also has to copy the argv data if I got that right,
    which is something it has to do always and not just when people are
    actually using the feature ;)

    I'm not saying I'm against this feature. I'd just really appreciate
    it if
    we could drop the good old "it doesn't matter if people are going to use
    it" non-arguments and instead provide a bit more info for people
    like me,
    who are not in the process-title-hacking business. I.e. what this is
    needed
    for an why this is needed in core. E.g. what Arvid mentioned, that
    this is
    useful when you are running many PHP-based daemons and want to
    distinguish
    them. That's the kind of stuff I'd like to see in the RFC.

    Regarding core/non-core. People mentioned that this is not
    implementable as
    an extension. That can be either solved by putting it into core or by
    adding the necessary API hook ;) [I'm not arguing which variant is
    better,
    just saying that not being implementable with current core does not mean
    that we can't make it implementable :)]

    Thanks,
    Nikita