Repository navigation
./configure output readability #595
Description
Since 1b1cd1c the ./configure output has gotten worse (I actually didn't mind the behavior before that commit). IMHO it's a step backwards.
This is what I see now:
{'variables': {'node_shared_http_parser': 'false',
'want_separate_host_toolset': 0, 'icu_small': 'false', 'node_use_mdb':
'false', 'target_arch': 'ia32', 'host_arch': 'ia32', 'v8_optimized_debug': 0,
'node_install_npm': 'true', 'openssl_no_asm': 0, 'uv_parent_path':
'/deps/uv/', 'v8_no_strict_aliasing': 1, 'node_shared_zlib': 'false',
'node_use_dtrace': 'false', 'python': '/usr/bin/python', 'v8_enable_gdbjit':
0, 'node_use_etw': 'false', 'v8_enable_i18n_support': 0, 'node_use_perfctr':
'false', 'v8_random_seed': 0, 'node_prefix': '', 'node_shared_openssl':
'false', 'uv_use_dtrace': 'false', 'node_shared_v8': 'false',
'node_use_openssl': 'true', 'uv_library': 'static_library', 'v8_use_snapshot':
'true', 'node_shared_libuv': 'false', 'node_tag': ''}, 'target_defaults':
{'libraries': [], 'default_configuration': 'Release', 'cflags': [],
'include_dirs': [], 'defines': []}}
Surely there is a better way? The keys aren't even in alphabetical order.
Activity
silverwind commentedon Jan 24, 2015
Could definitly be improved by proper nesting and sorting.
rvagg commentedon Jan 25, 2015
context: this was done because there was a new warning introduced for armv6 builds and it appears at the top of this block of output, hence the desire to shorten it because it generally overflows a standard terminal height.
I do agree that it is a bit of an eyesore now though, @trevnorris also commented about this
mscdex commentedon Jan 25, 2015
The warning can't be printed after the config option list?
rvagg commentedon Jan 25, 2015
not easily but you're welcome to refactor
silverwind commentedon Jan 25, 2015
Yeah, the reason for that change was that me and another user missed the warning as it scrolled off because of that long options list.
One possible solution could be to accumulate warnings and errors in a variable and print them at the end. Only issue I have with that would be that they are possibly taken out of context on which step they were generated on. Or alternatively, move the flags to the top (if possible).
silverwind commentedon Jan 25, 2015
After giving this more thought, I'd propose this:
pprintformThis will require a custom warn() function to be used throughout the file. I'll look into doing another PR for that.
bnoordhuis commentedon Jan 25, 2015
@silverwind You may want to base it off #585.
silverwind commentedon Jan 25, 2015
@bnoordhuis Thanks, that PR contains partly what I would've done. Will wait a day or two for that PR to resolve, and the submit my own.
bnoordhuis commentedon Jan 28, 2015
Fixed by 6ad236c.