I have simplified and improved the page
http://wiki.freepascal.org/Unicode_Support_in_Lazarus

I tried to move relevant information to the top and remove duplicate information.
For example there is a short "Usage" section with simple rules and then: "These rules make most code already compatible with Delphi." which is pretty much what most people need to know.

Please check.

Juha

* In section "Usage", "For string literals to work the source files must have UTF-8 encoding." - Add: "This is the default for source files written by Lazarus" - Otherwise people will begin to unnecessarily add these "$codepage utf8" directives.
* In section "CodePoint functions for encoding agnostic code" I am missing information on how "ch" is to be declared (in my test: ch:char and ch:string are working; ch:widechar does not).
* Good that the new page appears also in place of the old "Better Unicode Support..." page

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus-ide.org/pipermail/lazarus/attachments/20170513/aa25c4ac/attachment.html>

Am 13.05.2017 17:57 schrieb "Werner Pamler via Lazarus" <

http://wiki.freepascal.org/Unicode_Support_in_Lazarus I tried to move relevant information to the top and remove duplicate information. For example there is a short "Usage" section with simple rules and then: "These rules make most code already compatible with Delphi." which is pretty much what most people need to know.

In section "Usage", "For string literals to work the source files must

have UTF-8 encoding." - Add: "This is the default for source files written by Lazarus" - Otherwise people will begin to unnecessarily add these "$codepage utf8" directives.

Warning: the compiler does *not* consider the file as UTF-8 without a BOM or a set codepage!

Regards,
Sven
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus-ide.org/pipermail/lazarus/attachments/20170513/0a928895/attachment.html>

Werner Pamler wrote

In section "Usage", "For string literals to work the source files must have UTF-8 encoding." - Add: "This is the default for source files written by Lazarus" - Otherwise people will begin to unnecessarily add these "$codepage utf8" directives.

Actually that was one more duplicate information and in a wrong place. It is explained in "String Literals" section.
It is not part of simple rules because it happens automatically by default. Thanks for noticing.
It appears to be more difficult to remove old text than to write new text. Don't know why...

In section "CodePoint functions for encoding agnostic code" I am missing information on how "ch" is to be declared (in my test: ch:char and ch:string are working; ch:widechar does not).

I added variable types to the example.
I plan to improve the codepoint section more.
I will replace the page UTF8_strings_and_characters with a similar page that uses the LazUnicode encoding agnostic functions. Then the whole thing will be as Delphi compatible as possible.

Juha

Sven Barth wrote

Warning: the compiler does *not* consider the file as UTF-8 without a BOM or a set codepage!

Yes but that is a very misleading warning.
The whole thing works without a BOM better than with it because of the "double cheat". See "String Literals" section for details. It is counter-intuitive and most people can just ignore it.

With BOM it did not work for codepoints beyond BMP but that was fixed and backported to FPC 3.0.3, right?

Juha

Am 13.05.2017 19:45 schrieb "Juha Manninen via Lazarus" <

: With BOM it did not work for codepoints beyond BMP but that was fixed and backported to FPC 3.0.3, right?

Correct.

Regards,
Sven
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus-ide.org/pipermail/lazarus/attachments/20170513/6c824a9e/attachment-0001.html>