Hi,
>>> Can you describe what are the exact problem with
>>> current directory structure?
>>> It allows to keep SQLRDD code in one directory tree
>>> what seems to be clean solution for me better then
>>> dividing it into few directories in contrib dir.
>> I have to maintain rddsql specific hacks in
>> postinst.prg and packaging in winuni dir.
>> 'Maintain' means: Do it manually, develop script
>> for it, test it, all this just for the sake of
>> this exception. I especially don't like to
>> hard-wire contrib names in any non-contrib parts
>> of Harbour, but this is currently impossible.
>> Current solution also requires special chaining
>> of .hbc files, which makes them impossible to
>> move like the other .hbc files, it also requires
>> special logic for rddsql header inclusion as
>> opposed to standard HB_INC_DEPEND method. I could
>> go on.
>
> As I can see postinst.prg creates hbmk.cfg file which
> contains the following lines:
>
> libpaths=../contrib/%{hb_name}
> libpaths=../contrib/rddsql/%{hb_name}
> libpaths=../addons/%{hb_name}
> libpaths=../examples/%{hb_name}
>
> so I guess you want to remove the second one.
> But can you explain what above lines does?
> I was always wondering why I have it in hbmk.cfg
> when I do not have any of above directories in
> my Harbour installation.
It's a pending issue/question. Above will define
search path for .hbc files, so they are found
automatically without speccing full dir for them
in app-level make files.
I just have no idea how to solve that for *nix
systems. For non-*nix it's easy and already
working with above method.
> rddsql string is also used inside
> package/winuni/mpkg_win_uni_extra_copy.bat
> It's used to copy 'test' subdirectories and *.hbc files
> to some F:\hb\hb20\. I do not know what this code does
> and why it uses hardcoded F:\hb\hb20\. I only guess
> that it's bound in some way with generated hbmk.cfg.
>
> Maybe if you explain what hbmk.cfg does and why it
> uses such hardcoded directory structure then I will
> understand what you are trying to do. It should help
> all of us in adding new directories to SVN code.
> Now seems that we have some undocumented conditions
> which we should respect.
Partially related to hbmk.cfg above. The other
part is copying test and other stuff that comes
with contribs to winuni distribution.
This .bat file is not meant to be a "one-click"
winuni creator (as the included comments make it
obvious), it needs manual steps and adjustments
before usage. Anyhow this is irrelevant regarding
rddsql issue, I had/have to spend a large
amount of extra time and effort with such exceptions
either this file is on or off SVN. (you can now
see the simplified end-result of a longer struggle)
My time is finite and I hate to waste it on
such exceptions when there is no real reason
or benefit in them. In case of sdd*, they are
clearly separated by libname, so there is no
loss at all by moving them to main contrib
dir. They are simple libs depending on another
lib, no need to add extra complication by layering
the dirs as well.
Brgds,
Viktor
_______________________________________________
Harbour mailing list (attachment size limit: 40KB)
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
