Author Topic: Suggestion to assess software portability to OS/2  (Read 493 times)

Martin Iturbide

  • OS2World NewsMaster
  • Global Moderator
  • Hero Member
  • *****
  • Thank You
  • -Given: 407
  • -Receive: 79
  • Posts: 2408
  • Your Friend Wil Declares...
    • View Profile
    • Martin's Personal Blog
Suggestion to assess software portability to OS/2
« on: October 10, 2019, 09:57:28 pm »
Hi

I have a question for people with more experience porting and developing OS/2 software.
Are there any suggestions on how to assess how portable to OS/2 can an application be ??

For example, when I see an open source software, like OBS Studio, it is hard to find which libraries does it depends on to see if it is portable or not. Is there something you suggest to try to spot quickly how portable it is?

Regards
Martin Iturbide
OS2World NewsMaster
... just share the dream.

André Heldoorn

  • Hero Member
  • *****
  • Thank You
  • -Given: 119
  • -Receive: 19
  • Posts: 787
    • View Profile
Re: Suggestion to assess software portability to OS/2
« Reply #1 on: October 11, 2019, 12:01:29 am »
Is there something you suggest to try to spot quickly how portable it is?

Just try to build it and select the type of possible errors and warnings which you are willing to fix, while keeping in mind how important the software to be ported is(n't)? The compiler may produce all relevant error messages, like a missing library file or function, code errors, and so on. Without having to rely on documented requirements.

The more you want, for example un-Unixified code with an OS/2 icon file and a GUI, the harder it will get. Obviously an AUTOEXEC.BAT-sorter may be a better starting point than a wishlist with fatal requirements. Regarding your wishlist, I can see a few possible libraries which probably aren't available for OS/2 anyway. So then such a library is a better starting point than the wishlist item.
« Last Edit: October 11, 2019, 12:16:58 pm by André Heldoorn »

Dave Yeo

  • Hero Member
  • *****
  • Thank You
  • -Given: 11
  • -Receive: 149
  • Posts: 2170
    • View Profile
Re: Suggestion to assess software portability to OS/2
« Reply #2 on: October 11, 2019, 02:10:42 am »
I'd start by looking at the dependencies, here, https://github.com/obsproject/obs-studio/wiki/Install-Instructions under Linux. We do have most of them them, at least once QT5 is finished but I notice that the Linux port depends on OpenGl and Python 3. Also Pulse audio is a requirement, so audio support likely needs adding and a couple of other libraries that might be easy to port or not.
At this point can follow André's advice and try, including building the dependencies, or grep the source to see if 3D support is really required.

André Heldoorn

  • Hero Member
  • *****
  • Thank You
  • -Given: 119
  • -Receive: 19
  • Posts: 787
    • View Profile
Re: Suggestion to assess software portability to OS/2
« Reply #3 on: October 11, 2019, 01:57:50 pm »
I'd start by looking at the dependencies, here, https://github.com/obsproject/obs-studio/wiki/Install-Instructions under Linux.

The level or quality of the assessment is leading indeed. For the quickest assesment you could assume that there's no matching. documented OpenGL library for OS/2 yet, and stop. If I'm asking you to port something, because I don't have GCC installed, then you may stop right there. No OpenGL. A next level would be to check that the documentation is right. If so, then can OpenGL be ported. If so, quickly?

A starting point could be upgrading existing ports. A next step could be a port which requires a minor change (like replacing <win.h> by <os2.h>), or a port which requires a library you don't have installed yet (as you may have noticed, mister Ko tends to release several libraries, with source code and changes,  before releasing a new version of VLC, so you can assess what he's doing w.r.t. those required libraries).

If I'm requesting a port, then I'll take "Sorry, no OpenGL" as an answer. An example of all of the above may be Xpdf. You could try to port and release some of the EXEs of Xpdf easily, an assessment by compiling and an existing real request of me, but you can stop when Qt is involved and when it gets too difficult for a very quick "compile-only"port. Even for Elbert Pol.