On 09/27/2018 02:29 PM, Marty Kraimer wrote:
On 09/27/2018 07:18 AM, Ralph Lange wrote:
On Thu, 27 Sep 2018 at 09:53, Dirk Zimoch <[email protected]
<mailto:[email protected]>> wrote:
BTW: This leads to the question: If we have separate modules, why do
they all install their headers to the same directory "pv" instead
of one
directory for each module?
Because they are parts of the same bundle of software (the V4 parts we
just recently decided to call "PVA").
The members pvData, pvAccess, pvDatabase, pvClient, pvNT, ... are
closely related, released as a bundle and share the same prefix "pv",
which was selected to determine the name for the include subdirectory.
Cheers,
~Ralph
So then what should be done?
What I have been doing is two things:
1) All the copy modules will be accessed via include/pvcopy/*
I'm not terribly keen on our adding another include prefix directory,
especially for what is a fairly small API. I would rather that you just
rename the header file that pvDatabase provides, maybe call it pvCopy2.h
instead? That may be all you need to do at the moment.
2) All the copy code is in namespace epics::pvCopy
Okay.
I have it working but exampleCPP/exampleLink fails.
I found the problem is that
#include <pv/pvCopy.h>
appeared in pvaClent.h and in exampleLink.
When I removed them all build.
I have to do more more testing before I create a pull request for all of
pvaClientCPP exampleCPP and pvDatabaseCPP.
But what do I do about Ralph's comment above?
I don't think you have to do anything about Ralph's comment, he was
responding to a question from Dirk.
- Andrew