warning thrown

[kshahkshah]

Hi I'm testing out this handy gem, hope you're still considering active development?

Anyway, I'm getting a warning with Ruby 2.3.1:

In file included from /home/deploy/.rbenv/versions/2.3.1/include/ruby-2.3.0/ruby.h:33:0,
                 from /home/deploy/.ruby_inline/ruby-2.3.0/Inline_FuzzyTools__WeightedDocumentTokens_a329c30e0075fa4cebaee897338b8713.c:1:
/home/deploy/.rbenv/versions/2.3.1/lib/ruby/gems/2.3.0/gems/fuzzy_tools-1.0.0/lib/fuzzy_tools/weighted_document_tokens.rb: In function ‘cosine_similarity_fast’:
/home/deploy/.rbenv/versions/2.3.1/include/ruby-2.3.0/ruby/ruby.h:1254:22: warning: passing argument 1 of ‘rb_array_len’ makes integer from pointer without a cast [-Wint-conversion]
 #define RARRAY(obj)  (R_CAST(RArray)(obj))
                      ^
/home/deploy/.rbenv/versions/2.3.1/include/ruby-2.3.0/ruby/ruby.h:1044:36: note: in definition of macro ‘RARRAY_LEN’
 #define RARRAY_LEN(a) rb_array_len(a)
                                    ^
/home/deploy/.rbenv/versions/2.3.1/lib/ruby/gems/2.3.0/gems/fuzzy_tools-1.0.0/lib/fuzzy_tools/weighted_document_tokens.rb:49:39: note: in expansion of macro ‘RARRAY’
             for(i = 0; i < RARRAY_LEN(RARRAY(my_tokens)); i++) {
                                       ^
In file included from /home/deploy/.rbenv/versions/2.3.1/include/ruby-2.3.0/ruby.h:33:0,
                 from /home/deploy/.ruby_inline/ruby-2.3.0/Inline_FuzzyTools__WeightedDocumentTokens_a329c30e0075fa4cebaee897338b8713.c:1:
/home/deploy/.rbenv/versions/2.3.1/include/ruby-2.3.0/ruby/ruby.h:2010:1: note: expected ‘VALUE {aka long unsigned int}’ but argument is of type ‘struct RArray *’
 rb_array_len(VALUE a)
 ^

note that the gem installs fine w/in my Gemfile, this is upon require

[brianhempel]

Does is still work? i.e. does it not crash?

[kshahkshah]

@brianhempel still works! just tested it :) awesome

Only issue is creating the index, in a pry console e.g., spits back out the whole object which just floods the terminal with never-ending text, but that of course wouldn't be a problem in prod

[kshahkshah]

totally side note (sorry to cloud your GH issues) very fast still too! Hopefully I can come up with some way to Marshall and persist the model so I can retrieve and use it in background jobs processed by Sidekiq

[brianhempel]

There's a weight_function lambda that prevents Marshal from dumping the index. It looks like it is only used once and can just become a regular instance method of either WeightedDocumentTokens or TfIdfIndex. I abstracted too early.

Feel free to make that change and then it should Marshall fine.