Opened 7 years ago

Closed 6 years ago

#55112 closed defect (fixed)

poco build failure

Reported by: snhirsch Owned by:
Priority: Normal Milestone:
Component: ports Version:
Keywords: Cc:
Port: poco

Description

The underlying issue is not terribly clear from the build log. I made sure that my entire ports universe was updated and made several attempts at building (cleaned in between).

Attachments (2)

main.log.gz (270.5 KB) - added by snhirsch 7 years ago.
Build log
AtomicCounter.patch.txt (750 bytes) - added by snhirsch 7 years ago.
Patch for Mac OS X header issues

Download all attachments as: .zip

Change History (11)

Changed 7 years ago by snhirsch

Attachment: main.log.gz added

Build log

comment:1 Changed 7 years ago by Schamschula (Marius Schamschula)

It looks like poco is building a local version of sqlite3. However, clang is picking up the header files from the installed sqlite3 port, leading to the redefinition errors in your log. Interestingly, the portfile does not declare sqlite3 as a dependency...

Changed 7 years ago by snhirsch

Attachment: AtomicCounter.patch.txt added

Patch for Mac OS X header issues

comment:2 Changed 7 years ago by Schamschula (Marius Schamschula)

Port: poco added

comment:3 Changed 7 years ago by snhirsch

I was able to build Poco from the upstream sources, but when I went to build against it I ran into another problem. It was corrected with the attached patch from their issues tracker. I did notice compiler warnings referencing OSAtomicIncrement32 in the build log.

Would definitely be neater overall if the ports package worked. Let me know if there's anything else I can do to help?

comment:4 Changed 6 years ago by snhirsch

Just curious if any action is planned on this? We're having to build poco from patched upstream sources, which is sort of a PITA. Is there any further information you require from me?

comment:5 in reply to:  4 Changed 6 years ago by mf2k (Frank Schima)

Replying to snhirsch:

Just curious if any action is planned on this?

Not likely since the port has no maintainer. If you can supply a Portfile patch that fixes the problem, it is much more likely to get fixed in Macports.

comment:6 Changed 6 years ago by snhirsch

Blast... I don't really know much about the Macports build system, nor do I understand the failure mode. Based on my original log, can you give me a quick hand-waving explanation of what's going astray and where? Also, what is 'clang'? Not familiar with that term.

Disclaimer: I am a reasonably experienced C/C++ coder in the Linux environment, but know next to nothing about the Mac OS X system (besides its lamentable lack of support for Valgrind, but that's a different story).

comment:7 Changed 6 years ago by arjanvandervelde (Arjan van der Velde)

i submitted a pull request to update to 1.8.0.1 (current upstream version) and fix the sqlite dependency; by default the port builds its own zlib, pcre, sqlite and so on libraries; i updated the Portfile to make it depend on the actual ports for those libraries (by adding the option '--unbundled' to configure). Also, it now builds both shared AND static libraries (that's why I ultimately needed this fix myself). Hope this helps. See: https://github.com/macports/macports-ports/pull/1073

comment:8 Changed 6 years ago by arjanvandervelde (Arjan van der Velde)

In 2dff4bc8f6940bd1de2b94993cf49bd79e112dd8/macports-ports:

poco: fix build issue and bump version to 1.8.0.1

Bumped the version of this broken port to 1.8.0.1. This also this fixes (or supersedes) the build error in #55112.

comment:9 Changed 6 years ago by mf2k (Frank Schima)

Resolution: fixed
Status: newclosed
Note: See TracTickets for help on using tickets.