Subject : Re: Long time no see
VICTOR : Hi Chong Hwa,
Nice to see you reviving Clipper...Clipper...Clipper.
It's been a long time :) I hope you're well.
CCH: Well, it's been more than 10 years :-)
VICTOR : Yep. I'm glad you remember.
VICTOR : If you're interested, I invite you to join Harbour's developer list,
it'd be great to see you there.
CCH: TQ and I am honoured by your invitation.
However, work commitments does not me to do so :-)
VICTOR : I can understand that, you're welcome any time in the future.
CCH: I would love to publish an article from you showing how you switched
from Clipper to Harbour. I am OK with xHarbour but I must admit I am a little lost on how to do it with Harbour.
From Clipper to Harbour - Viktor Szakáts
From a user's POV they are probably the same. Maybe there are some aspects of xhb which I don't know and could make the process different, but normally it's just a recompile and relink, feature-wise Harbour is on par with xhb, and a plus Harbour is cleaner, more robust, more portable and faster, it also has an xhb.lib which aims to provide compatibility with
features we've opted not to include in the core product.
I've just finished a new hbmk (Harbour Make) utility this weekend.
My transition was a long, gradual process, while keeping continuous Clipper releases and focusing on minimizing any risks on the user side, while maturing the Harbour release versions.
Also, the process was two fold:
1) Transition to Harbour for Windows.
2) Transition to be multi platform in general, allowing
easy move to any other platforms in the future.
This mostly comes down to file system / OS environment
differences.
Here are the main stages (and the devils are in the details :):
1) Departing from DOS/Windows concepts like case-insensitive filenames, drive letters and any other possible platform dependencies, like using some specific external tools. This was done by keeping Clipper compatibility, where this was impossible it was added using conditional code.
2) Some libs/tools which didn't have a Harbour/portable version, had to be replaced by new parts and sometimes concepts. One such is Telepath(y), which got replaced by curl, then hbcurl.
3) Adjusting all own .c functions for Harbour Windows version, using conditional code.
4) Added simple make system for Harbour Windows versions.
5) Created first running Harbour version for Windows and started testing locally plus a few clients.
6) After filling all missing bits in Harbour, there was time for the show to begin by creating the first official Harbour builds of the apps. Points below took 3-6 months depending on how we count it.
7) This quickly turned out to be a success since as apps were proved to run very stable, now the main work could begin by now being able to drop all Clipper heritage.
8) Created new local build systems for all parts, this time keeping in mind Linux/Darwin compatibility.
9) Started optimizing speed by refining the compiler choice and Harbour and local app build options.
10) After removing dependency on Clipper, the code could be moved from Clipper compatibility mode to Harbour native mode. This allowed to drop many functions and replace them with Harbour functions, tools and libs. This solved some rare problems with the apps, caused by bugs in local .c code. In fact I've reduced local .c code to an absolute minimum. This stage also resulted in noticeable speed up in certain points of the application.
11) The apps were now ported to Linux/Darwin with minimal effort.
12) Now fun part could be started, really start to exploit the possibilities of the new platforms and Harbour. I'm doing this since months and it's really a breath of fresh air.
Since I'm a main contributor to Harbour, all this was accompanied by parallel Harbour development, fixing incompatibilities, some bugs, finding all missing points and filling the gaps, and creating stable releases.
In the last few years the main areas to fill were TBROWSE, some multi-platform file handling stuff for the most part, and some stuff related to codepages. Currently using the latest SVN, which is practically release quality.
I hope that helps or is useful in any ways.
Brgds,
Viktor Szakáts
No comments:
Post a Comment