compiling boost/iostreams with external lzma links to wrong lzma library about iostreams

[jeking3]

According to boost build sources, you want to define:

jking@ubuntu:~/boost/tools/build$ grep -r LZMA .
./src/tools/lzma.jam:# variables LZMA_LIBRARY_PATH, LZMA_NAME, and LZMA_INCLUDE will

jking@ubuntu:~/boost/tools/build$ grep -r BZIP2 .
./src/tools/bzip2.jam:# variables BZIP2_LIBRARY_PATH, BZIP2_NAME, and BZIP2_INCLUDE will

jking@ubuntu:~/boost/tools/build$ grep -r ZLIB .
./src/tools/zlib.jam:# variables ZLIB_LIBRARY_PATH, ZLIB_NAME, and ZLIB_INCLUDE will

Give those a try and see if they work.

[swatanabe]

AMDG

On 11/28/2018 07:08 PM, James E. King III wrote: According to boost build sources, you want to define: ``` ***@***.***:~/boost/tools/build$ grep -r LZMA . ./src/tools/lzma.jam:# variables LZMA_LIBRARY_PATH, LZMA_NAME, and LZMA_INCLUDE will <snip> ``` Give those a try and see if they work.

It shouldn't make any difference. The other names are still retained for backwards compatibility. In Christ, Steven Watanabe

[jeking3]

I assume there is a log file one could look at to see the results of the resolution of those libraries?

[KBentley57]

I was under the impression that those names would work fine, so I left them in the older style. I tried it anyway for the sake of completeness, but it didn't seem to matter. There's nothing of value in the config.log file that I can see, but it's attached, along with the build output and the installation output. I'm going to follow up with a minimal test case, if either of you are interested in trying to replicate the build. Thanks for the help. Kyle

…

On Wed, Nov 28, 2018 at 8:46 PM James E. King III ***@***.***> wrote: I assume there is a log file one could look at to see the results of the resolution of those libraries? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#73 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AQsOCzGPjkjlyEUm4-fvzyjF2-irHNt5ks5uz0p6gaJpZM4Y4d2a> .

[KBentley57]

I thought emailing the logs would attach them here, but I suppose that's how little I know about github! Anyway, I've created a little repo to aid in the diagnosis. If you clone it, you should be able to run the test, and reproduce the linking error.

https://github.com/KBentley57/boost-iostreams_test

[KBentley57]

Just another note, I have tried building it without the CMake baggage, using the same inputs to bootstrap and b2. It turns out the same.

[KBentley57]

Update - This no longer seems to be the behavior as of 1.70.0. I'm closing this issue, as all are now correctly identified.