Self referencing example #56
Commits on Oct 23, 2020
Commits on Oct 27, 2020
Commits on Oct 28, 2020
-
-
-
-
-
-
Merge pull request #8 from jakewilliami/dev
Updated version number
-
-
-
-
-
Merge pull request #10 from jakewilliami/dev
Added OS BSD to .travis.yml
-
-
Merge pull request #11 from jakewilliami/dev
Updated version number
-
Commits on Nov 1, 2020
-
-
-
Merge pull request #16 from jakewilliami/dev
Implemented floating point spelled out (closes #14)
-
-
Merge pull request #17 from jakewilliami/dev
Began work on complex numbers (addresses #15)
-
Merge pull request #18 from jakewilliami/dev
Removed 1.0 from travis testing, as DecFP requires 1.3 or greater
Commits on Nov 28, 2020
Commits on Dec 30, 2020
-
-
Merge pull request #21 from jakewilliami/compathelper/new_version/202…
…0-11-28-00-39-26-535-3281016136 CompatHelper: add new compat entry for "DecFP" at version "1.1"
Commits on Jan 12, 2021
-
-
Merge pull request #23 from jakewilliami/universal
Refactored for setup for multiple language support
-
-
-
-
-
-
-
-
-
-
Commits on Jan 13, 2021
Commits on Jan 14, 2021
-
Removed kwarg from exported function and replaced with language varia…
…nt; updated docs accordingly
-
-
-
Merge pull request #29 from jakewilliami/universal
Default language to system locale (suggestion by @rmsrosa)
Commits on Jan 18, 2021
-
Merge pull request #33 from jakewilliami/universal
Allowed english SpelledOut to parse big numbers in scientific notation
-
-
Merge pull request #34 from jakewilliami/universal
Added formatting dependency to docs-specific dependencies
Commits on Jan 19, 2021
Commits on Jan 21, 2021
Commits on Jan 24, 2021
-
Merge pull request #35 from jakewilliami/compathelper/new_version/202…
…1-01-19-01-24-31-237-1730833285 CompatHelper: add new compat entry for "Formatting" at version "0.4"
-
-
-
Commits on Jan 25, 2021
Commits on Jan 27, 2021
-
-
-
-
Merge pull request #45 from jakewilliami/fallback
Added fallback to error for bad input type (closes #41)
-
-
Merge pull request #47 from jakewilliami/rational
Added parsing for rational numbers (closes #39)
-
Added method handling irrational numbers (closes #38, but open questi…
…on addressed in #48)
-
Merge pull request #49 from jakewilliami/irrational
Added method handling irrational numbers (closes #38, but open questi…
-
-
Commits on Aug 21, 2021
Commits on Oct 15, 2021
-
-
Merge pull request #51 from rick2047/docstring
correct syntax for british spell out in docs
Commits on Jun 1, 2022
Commits on Jun 2, 2022
Commits on Jun 21, 2022
-
-
-
Merge pull request #52 from Gabrielforest/master
Implement language support for Portuguese
Commits on Jun 26, 2022
-
-
-
-
-
Instead of creating IOBuffers within each function, pass around IOBuffers to modifying functions and only take from them at the end of the process! This vastly reduces the allocations as we are no longer creating a new IOBuffer within each function. However, it is still not as fast as it was (but its memory usage is better).
-
-
Commits on Jun 27, 2022
-
-
-
Better string manipulation and use buf for niche convert funcs
Instead of using a weird `reverse` method for string manipulation in printing of ordinals, we use a util function. Also, use `IOBuffer` where practical for alt convert methods
-
-
-
Merge pull request #53 from jakewilliami/optimisation
Optimise English implementation
Commits on Sep 22, 2022
-
WIP: begin example project using spelled_out
A self-referencing, spelled-out sentence such as "This sentence has ten 'a's and four 'b's and twelve 'c's and ..." would be a cool example project to make
-
Implement brute force and continue clever solution
Implement Doov's brute force idea, and continue work on my attempt at a somewhat efficient solution (though I think that Doov's idea is perhaps the best way, as I can see my solution becoming quite complex)
Commits on Jan 9, 2023
-
Initial commit for working self-referencing example
Previous commits working on this did not finish computing, as they would form cycles. The way around this is to use some level of randomisation. This commit implements a working solution. Optimisation and comments to follow.
-
Set diffs to modify previous countmap in self-referencing example
When we calculate the diffs from previous to current countmaps in self-referencing example, we previously created a new dictionary. As we never use the previous countmap after this point, we can simply modify it! This seems to speed things up immensely: Without mutation of previous: julia> @Btime construct_true_pangram(); 8.100 s (65276305 allocations: 4.59 GiB) With mutation of previous: julia> @Btime construct_true_pangram(); 2.992 s (26294381 allocations: 1.67 GiB) Next steps for optimisation would be to reuse the same buffer for pangram construction. -
Some code clean up within self-referencing example
In this commit, I dynamically determined when we were at the second-to-last or last item in our alphabet, as well as some other miscellaneous clean up/improvement. I had planned to only keep track of one IOBuffer in this commit, but after some benchmarking, this did not seem to improve performance: With one IOBuffer: julia> for _ in 1:20 @Btime construct_true_pangram(); end 50.843 s (409455124 allocations: 26.26 GiB) 765.545 ms (6216088 allocations: 408.26 MiB) 3.035 s (24595631 allocations: 1.58 GiB) 33.922 s (273593730 allocations: 17.55 GiB) 2.409 s (19406731 allocations: 1.24 GiB) 1.898 s (15389028 allocations: 1010.77 MiB) 11.832 s (95874131 allocations: 6.15 GiB) 44.362 s (357949407 allocations: 22.96 GiB) 2.459 s (19967197 allocations: 1.28 GiB) 58.991 s (475764808 allocations: 30.52 GiB) 1.158 s (9389309 allocations: 616.70 MiB) 25.636 s (207005638 allocations: 13.28 GiB) 51.847 s (418013672 allocations: 26.81 GiB) 31.430 s (253300737 allocations: 16.25 GiB) 40.973 s (327232985 allocations: 20.99 GiB) 4.422 s (35438747 allocations: 2.27 GiB) 253.188 ms (2041985 allocations: 134.09 MiB) 36.172 s (292155055 allocations: 18.74 GiB) 49.306 s (396551677 allocations: 25.44 GiB) 55.549 s (446664880 allocations: 28.65 GiB) With IOBuffer constructed in each construct_pangram call: julia> for _ in 1:20 @Btime construct_true_pangram(); end 5.503 s (47207580 allocations: 3.00 GiB) 7.020 s (60222956 allocations: 3.82 GiB) 12.165 s (103483280 allocations: 6.57 GiB) 3.002 s (25651341 allocations: 1.63 GiB) 5.414 s (46572408 allocations: 2.96 GiB) 5.389 s (45930841 allocations: 2.92 GiB) 5.182 s (44124955 allocations: 2.80 GiB) 9.869 s (85081002 allocations: 5.40 GiB) 2.006 s (17076024 allocations: 1.08 GiB) 646.730 ms (5628166 allocations: 365.71 MiB) 666.316 ms (5935178 allocations: 385.65 MiB) 1.553 s (13790449 allocations: 896.04 MiB) 11.421 s (101638768 allocations: 6.45 GiB) 199.693 ms (1800406 allocations: 117.00 MiB) 20.385 s (180114986 allocations: 11.43 GiB) 838.932 ms (7439715 allocations: 483.56 MiB) 868.218 ms (7668002 allocations: 498.32 MiB) 9.911 s (87764372 allocations: 5.57 GiB) 890.219 ms (7936633 allocations: 515.73 MiB) 11.319 s (100671495 allocations: 6.39 GiB I will have to investigate this a little further, as I don't think it should make such a difference. Unfortunately, it's a little tricky to benchmark with it being random! -
Remove initial state from static chars in self-referencing example
While it is true that certain characters in this sentence are never used in spelling out numbers, creating an initial state accounting for these characters has not proven useful, and hence this commit removes this logic. I'm not entirely sure why, but for some reason, having this initial state considerably slows down the programme.
-
Construct IOBuffer only once in self-referencing example
Despite notes made in 7849a1c, after testing again it appears that constructing IOBuffer only once, at the start of the main function, provides a slight performance improvement (at least in terms of memory, and perhaps negligible difference in speed, but as previously noted, it is difficult to benchmark random functions).
-