Closed kbarbary closed 6 years ago
Merging #730 into master will decrease coverage by
0.38%
. The diff coverage is95.49%
.
@@ Coverage Diff @@
## master #730 +/- ##
==========================================
- Coverage 81.57% 81.19% -0.39%
==========================================
Files 39 39
Lines 3963 3961 -2
==========================================
- Hits 3233 3216 -17
- Misses 730 745 +15
Impacted Files | Coverage Δ | |
---|---|---|
src/deterministic_vi/elbo_kl.jl | 51.19% <ø> (ø) |
:arrow_up: |
src/deterministic_vi/elbo_objective.jl | 99.43% <ø> (ø) |
:arrow_up: |
src/deterministic_vi/ElboMaximize.jl | 100% <100%> (ø) |
:arrow_up: |
src/mcmc/mcmc_infer.jl | 47.32% <100%> (ø) |
:arrow_up: |
src/model/light_source_model.jl | 6.89% <100%> (-51.73%) |
:arrow_down: |
src/deterministic_vi/source_brightness.jl | 100% <100%> (ø) |
:arrow_up: |
src/PSF.jl | 98.56% <100%> (ø) |
:arrow_up: |
src/model/param_set.jl | 15% <100%> (ø) |
:arrow_up: |
src/model/wcs_utils.jl | 100% <100%> (ø) |
:arrow_up: |
src/deterministic_vi/elbo_args.jl | 100% <100%> (ø) |
:arrow_up: |
... and 6 more |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update a06adfc...e0f4e15. Read the comment docs.
Great, glad we'll be ready for julia 1.0.
This updates all parameteric method definitions to use the new
where
syntax in Julia 0.6 (I believe required in 0.7).Along the way, I made the style more uniform and conventional by always using
T
for a numeric type rather thanNumType
and always using 4-space indents. (Reindenting is the reason for the large diff.)This also has two bugfixes:
In
src/mcmc/mcmc_infer.jl
: Sometimes the@sprintf
calls were resulting in a string with an embedded NULL, causing an error when passed toLog.info()
. I believe (but am not certain) that this happens when the format code is not long enough for the value. I widened the format codes.In
src/model/light_source_model.jl
: We were getting errors from JLD/HDF5 that might have been due to from callingload_prior()
from threads (load_prior()
does I/O with JLD). I replaced theload_prior()
function. It now returns a copy of the prior loaded on module import.load_prior_init()
does the I/O and is now called only on module import.