#17982 closed defect (fixed)
libiconv: universal LP64 fixes
| Reported by: | jmroot (Joshua Root) | Owned by: | ryandesign (Ryan Carsten Schmidt) |
|---|---|---|---|
| Priority: | Normal | Milestone: | |
| Component: | ports | Version: | 1.7.0 |
| Keywords: | Cc: | MarcusCalhoun-Lopez (Marcus Calhoun-Lopez) | |
| Port: | libiconv |
Description
Attached patches correct libiconv's config.h when building for both 32- and 64-bit.
Attachments (3)
Change History (15)
Changed 15 years ago by jmroot (Joshua Root)
| Attachment: | libiconv-lp64.diff added |
|---|
Changed 15 years ago by jmroot (Joshua Root)
| Attachment: | patch-config.h.diff added |
|---|
comment:1 Changed 15 years ago by MarcusCalhoun-Lopez (Marcus Calhoun-Lopez)
| Cc: | mcalhoun@… added |
|---|
comment:2 Changed 15 years ago by MarcusCalhoun-Lopez (Marcus Calhoun-Lopez)
Using #17972, there are a few more changes in the 32/64 build processes.
I do not know if they matter.
Changed 15 years ago by MarcusCalhoun-Lopez (Marcus Calhoun-Lopez)
| Attachment: | build_diff_32_64.diff added |
|---|
comment:3 Changed 15 years ago by ryandesign (Ryan Carsten Schmidt)
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
Thank you both. I applied a combination of your fixes in r45447.
comment:4 Changed 15 years ago by jmroot (Joshua Root)
Unfortunately configuring libiconv now fails when the universal variant is not set, and also when building universal but not 64-bit. That long if statement in my patch was there for a reason.
comment:5 Changed 15 years ago by ryandesign (Ryan Carsten Schmidt)
| Resolution: | fixed |
|---|---|
| Status: | closed → reopened |
Hum... I didn't understand why the patch had to be restricted to the universal variant, and I thought I had tested without it...
comment:6 Changed 15 years ago by ryandesign (Ryan Carsten Schmidt)
Ok, I see the problem is that the patch does not apply unless you build with 64 bits.
I wanted to do the fix all the time, not just in the universal variant, based on Joshua's comment in #16007 that some systems may want to build 64-bit all the time. I guess I should convert the patch into an ed script, as per Joshua's comment in #17042.
comment:7 Changed 15 years ago by ryandesign (Ryan Carsten Schmidt)
Or have you changed your mind on that idea?
comment:8 follow-up: 10 Changed 15 years ago by jmroot (Joshua Root)
Daniel didn't like the idea of adding a dependency on ed in #17090. Maybe it's not a big deal.
comment:9 Changed 15 years ago by jmroot (Joshua Root)
Also, in #16007 it was broken on 64-bit. In this case it's only broken when building for two different word lengths at once.
comment:10 Changed 15 years ago by ryandesign (Ryan Carsten Schmidt)
Replying to jmr@…:
Daniel didn't like the idea of adding a dependency on ed in #17090. Maybe it's not a big deal.
I think using ed is fine because ed is in /bin/ed while patch is in /usr/bin/patch, suggesting that ed is a lower-level utility than patch, which we already require, so ed should be on at least as many systems as have patch.
comment:11 Changed 15 years ago by ryandesign (Ryan Carsten Schmidt)
| Resolution: | → fixed |
|---|---|
| Status: | reopened → closed |
Now using an ed script in r45487. Increasing revision again because I got PTRDIFF_T_SUFFIX reversed in my first attempt.
Cc Me!