BUG: hrsh2csv fix for Intel

[css@…]

hrsh2csv uses a configure-time setting to account for the endian ordering. Fix the port so that the macro is not disabled when building on an Intel platform. The port currently fails to process PDB files, as it expects a little endian platform.

[css@…]

Create a powerpc platform variant for the removal of byte order switching

[css@…]

Fixed in r28197

[ryandesign@…]

r28197 broke the build on PowerPC, because now it's trying to patch the Makefile before it's been extracted. It also broke the portindex, when indexing occurs on a PowerPC machine (which it currently does).

I was also going to suggest looking explicitly at the endian variable, rather than using the powerpc platform selector.

I recommend applying the attached patch hrsh2csv-endian.diff. I hope this fixes it. It is untested, except insofar as the endian check was copied out of the working wine port.

[css@…]

That fix seems to work fine. I tested it building on my MacBook Pro, but I no longer have access to a PowerPC Mac.

[css@…]

What about this approach of using pre-configure to keep the modification within a platform variant?

[ryandesign@…]

jmpp has committed a fix for the issue in r28318. I still would have preferred my fix in hrsh2csv-endian.diff because my fix handles all big-endian architectures, not just powerpc.

[jmpp@…]

Hey Ryan! Had completely forgotten about this ticket, please don't feel like I ignored it ;-) Feel free to commit your patch if you feel it does the job better than mine!

-jmpp

[blb@…]

Was there ever a consensus on this issue? Also, should this port list universal_variant no for now?

[css@…]

Replying to blb@…:

Also, should this port list universal_variant no for now?

Yes, it probably should, as it's not currently a universal-friendly port. As for the endian issue, I like the most general solution available. I can adjust the Portfile accordingly.

[anonymous]

Milestone Port Bugs deleted

[jmr@…]

This appears to be long since fixed.