Re: [pcre-dev] EBCDIC many faces

Top Page
Delete this message
Author: ph10
Date:  
To: Ze'ev Atlas
CC: Pcre Exim
Subject: Re: [pcre-dev] EBCDIC many faces
On Wed, 18 Oct 2017, Ze'ev Atlas wrote:

> PhilI have encountered this issue.  IBM did really bad job in creating
> the EBCDIC, especially as far as pattern matching is concerned.  When
> they adapted EBCDIC to various languages they moved these symbols
> around like a whirpool.  For example square bracket occupies 3
> different possitions in 3 major variations of EBCDIC, 037 - North
> America, 285 - United Kingdom and 1047 - the C compiler default.  One
> cannot make it up, it is reality.I need your guidance.  I think that I
> remember that one may make PCRE use special character tables at will. 
> If this is correct, I may instruct the users to use their version of
> EBCDIC while still allow the C compiler to use its default.  Do you
> think it will work, or should I just instruct the users to forgo the
> default and compile under their own specific variation (in tat case I
> cannot distribute binaries and not all users have access to the C
> compiler... don't ask)If using special character tables is an option,
> please guide me to where is that discussed.Thank youZe'ev Atlas


Ze'ev,

The use of special tables is described in the section called LOCALE
SUPPORT in the pcre2api documentation. It relies on functions such as
isdigit() etc "doing the right thing" in your environment. (I'll send
you more details off-list.)

Philip

--
Philip Hazel