diff options
author | Sebastian Rasmussen <sebras@gmail.com> | 2016-06-14 02:08:28 +0200 |
---|---|---|
committer | Sebastian Rasmussen <sebras@gmail.com> | 2016-06-14 02:10:24 +0200 |
commit | 9c9fe1a6e373ab174667314d5be29c6822894f54 (patch) | |
tree | d3649b24e3e7afee05c5cb71b4e25aff66a8c3a5 /source/fitz/strtof.c | |
parent | 4d7bf58d137d884e4eb0e81fd576031f069105cb (diff) | |
download | mupdf-9c9fe1a6e373ab174667314d5be29c6822894f54.tar.xz |
Fix typos in various parts of the code.
Diffstat (limited to 'source/fitz/strtof.c')
-rw-r--r-- | source/fitz/strtof.c | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/source/fitz/strtof.c b/source/fitz/strtof.c index a6f2f983..2020dafe 100644 --- a/source/fitz/strtof.c +++ b/source/fitz/strtof.c @@ -23,7 +23,7 @@ 4) Round x to a float using rounding mode 'to even'. Step 1) is always lossless as the strtof_fp_t's significand can hold a 9-digit integer. - In the case |dexp| <= 13 the cached power is exact and the algoritm returns + In the case |dexp| <= 13 the cached power is exact and the algorithm returns the exactly rounded result (with rounding mode 'to even'). There is no double-rounding in 3), 4) as the multiply/divide uses 'round to odd'. @@ -197,7 +197,7 @@ diy_to_float(strtof_fp_t x, int negative) /* We have 2^32 - 2^8 = 0xffffff00. */ else if (x.e > -158) { - /* x is greator or equal to FLT_MAX. So we get a normalized number. */ + /* x is greater or equal to FLT_MAX. So we get a normalized number. */ result = (uint32_t) (x.e + 158) << SP_SIGNIFICAND_SIZE; result |= (x.f >> 8) & SP_SIGNIFICAND_MASK; @@ -239,7 +239,7 @@ diy_to_float(strtof_fp_t x, int negative) else if (x.e == -181 && x.f > 0x80000000) { /* x is in the range (0.5,1) * 2^-149 so it rounds to the smallest - denormal. Cant't handle this in the previous case as shifting a + denormal. Can't handle this in the previous case as shifting a uint32_t 32 bits to the right is undefined behaviour. */ result = 1; } |