[fpc-devel] Breaking change in FPC 2.6.1

Sven Barth pascaldragon at googlemail.com
Tue May 1 22:05:08 CEST 2012


On 01.05.2012 22:03, Sven Barth wrote:
> On 01.05.2012 20:15, Giuliano Colla wrote:
>> Il 01/05/2012 17:08, Michael Van Canneyt ha scritto:
>>>
>>>
>>> On Tue, 1 May 2012, Hans-Peter Diettrich wrote:
>>>
>>>> Michael Van Canneyt schrieb:
>>>>
>>>>> Well, then they'll have to live with the warning.
>>>>>
>>>>> And this is the point of having the warning in the first place. Make
>>>>> people aware of a coming change.
>>>>
>>>> As already mentioned in this thread, a mere hint about "may change
>>>> somehow, in the next version" is of no real use. IMO "deprecated"
>>>> means to the user that he should change his code *now*, in
>>>> anticipation of the coming change. This obviously is not the case
>>>> with the breaking change of the bookmark type, where *no* workaround
>>>> exists in the current release.
>>>
>>> Instead of repeating the remark, suggesting a solution would be useful.
>>>
>>> Criticism is easy. Coming up with solutions obviously much less so.
>>>
>> I never used TDataset.Bookmark, and I'm not likely to use it in future,
>> so my attitude is that of an external observer, which might be
>> completely unaware of implications, but, as a general rule, wouldn't it
>> be wise to surround with an {$mode Delphi} all changes made to comply
>> with Delphi whims, to satisfy those who need Delphi compatibility, and
>> keep for the rest of the users things as stable and consistent as
>> possible, avoiding whenever possible to break compatibility with
>> existing code?
>
> You can not "surround" code with {$mode Delphi}. It's a unit global
> option and also it only changes syntax and semantic handling in the
> compiler not code in the units.

Also we are talking about three kinds of Delphi compatible here:
* pre-Delphi 2007 for which used String
* Delphi 2007 which used Pointer
* Delphi 2009 and newer which uses TBytes

This is another reason why using something like "{$mode delphi}" would 
be useless here.

Regards,
Sven




More information about the fpc-devel mailing list