RE: [PHP-QA] Re: [PHP-DEV] Re: [PHP-QA] Naming changes.
| From: | James Moore | Date: | Thu, 28 Dec 2000 17:56:18 +0000 |
| Subject: | RE: [PHP-QA] Re: [PHP-DEV] Re: [PHP-QA] Naming changes. | ||
| References: | 1 | Groups: | php.dev php.qa |
| Request: | Send a blank email to php-dev+get-42452@lists.php.net to get a copy of this message | ||
Another option we should look at is having the following in configure
--compat-mode /* depreciated functions issue notices */
--strict-mode /* depreciated functions dont work at all */
then when developing new stuff (ie on dev servers) people should use
strict-mode but on production servers --compat-mode should be used
(especially by ISP's and the like)
James
> -----Original Message-----
> From: Richard Lynch [mailto:richard@zend.com]
> Sent: 28 December 2000 17:45
> To: PHP Quality Assurance Team Mailing List; PHP Developers Mailing List
> Cc: Zeev Suraski; James Moore; Sterling Hughes
> Subject: Re: [PHP-QA] Re: [PHP-DEV] Re: [PHP-QA] Naming changes.
>
>
> > Actually, I think that shattering compatibility like this should be a
> minor
>
> IMHO:
>
> If the functions that used to work are issuing Warnings in the first phase
> of change, I'd also say that's worth a 4.1 Otherwise, web-sites using
> error_reporting=15 will break all over the place on a minor revision, and
> that's bad... Actually, even with a 4.1 that would be bad...
>
> If the DOCUMENTATION is changing first, along with both function names
> silently "working", and then in a *later* revision the Warnings start
> appearing, followed by escalating error signals (as outlined by James, I
> think) minor revisions would be okay...
> But it would probably be good to not escalate every minor revision -- More
> like 2 or 3 minor revisions per escalation to give people time to
> adjust not
> only their code, but to the "plan"
>
> Perhaps all renamed functions should have an auto-generated link to a web
> page explaining the procedure, whatever it is decided to be, and
> the list of
> planned name changes with a time-table or revision schedule... If that
> could be worked into the macro[s], it would probably help a lot.
>
> I guess I'm suggesting the following "phases" for James' plan:
>
> Phase I
> 1. Documentation notes both function names, clearly marking the
> deprecated one, with a link to a page describing the scheduled changes and
> list of functions to change.
> 2. Both functions work with no warning/error/etc
>
> Phase II, at least 2 or 3 minor revisions later
> 1. Documentation notes both function names, clearly marking the
> deprecated one with a note about the warning message, and the same link as
> above.
> 2. Deprecated function issues the lowest level Warning possible.
>
> Phase III and following, at least 2 or 3 minor revisions after previous
> Phase
> Same as above, but one level higher in the error system.
>
> Final Phase -- The deprecated function should remain in the
> manual another 2
> or 3 revisions, and finally be phased out.
> The *new* function names should continue to point to the page
> describing the
> changes that occurred and list of functions that changed at least through
> and including the next *major* revision.
>