yamcha @ 0.33 - new port

[humem (humem)]

This is a Portfile for YamCha, Yet Another Multipurpose CHunk Annotator.

[humem (humem)]

Portfile.2 enables a universal build and make check as well as the tinysvm port I submitted.

[ryandesign (Ryan Carsten Schmidt)]

Thanks.

[ryandesign (Ryan Carsten Schmidt)]

Added yamcha in r78011 with these changes:

• fixed double slash in master_sites
• removed universal variant stuff; not necessary in a project like yamcha with a standard autoconf-based configure script; it only failed to build universal because the configure script erroneously overrides our CFLAGS and CXXFLAGS; I added a patch to fix that
• changed configure.args-append back to just configure.args; configure.args is empty by default; it's there for the portfile to set as desired (not necessarily the case when using a portgroup, but this portfile isn't using a portgroup)
• indicated version of LGPL in the license field
• corrected case of TinySVM dependency
• enabled building of shared library (why had you disabled it?)
• in order to get the shared library to build for the right architectures, since the configure script was made with a too-old version of autoconf, brought back "use_autoreconf yes"; this also makes the --mandir argument unnecessary
• fixed tests to run correctly by specifying correct location for svm_learn in --with-svm-learn argument
• used lindex in livecheck.url definition

[humem (humem)]

Replying to ryandesign@…:

• enabled building of shared library (why had you disabled it?)

I totally thought I should disable building of shared library just as the TinySVM port, but I realize I was wrong.

• used lindex in livecheck.url definition

I wonder whether livecheck.url definition is not necessary in this port.

[ryandesign (Ryan Carsten Schmidt)]

Replying to hiroshi.umemoto@…:

I wonder whether livecheck.url definition is not necessary in this port.

It depends on your intention. Currently, the livecheck is looking for new distfiles in the download directory. If you did not specify livecheck.url, it would look for them in the homepage. Currently, for yamcha, either one works. I tend to like to use directory listings; they seem less likely to fail in the future. Also, in this case, the directory listing is only about 2K of data while the homepage is about 32K of data, so the directory listing should be faster to fetch.