Changes

Summary

1. add void to lowerBound() and upperBound() declaration, which is a workaround to the warning message (details)
2. update thesis (details)
3. comment out unused variable declaration (details)
4. fix dangling reference after delete (details)
5. Made sure a variable is always initialized and removed some trailing whitespace. (details)
6. Order of evaluation between arguments of a function is not left-to-right. Or right-to-left for that matter, it is unordred. (details)
7. Removed SizeofExpr::expr and AlignofExpr::expr, expressions that would be stored there are wrapped in TypeofType and stored in the type field. Some special cases to hide the typeof in code generation were added. In addition, initializer length is calculated in more cases so that the full type of more arrays is known sooner. Other than that, most of the code changes were just stripping out the conditional code and checks no longer needed. Some tests had to be updated, because the typeof is not hidden in dumps and the resolver replaces known typeof expressions with the type. The __extension__ case caused some concern but it appears that just hides warnings in the expression which no longer exists. (details)
8. Removed some unneeded code that could have gone last commit. (details)
9. Add separate compilation discussion (details)
10. Update thesis (details)
11. Fix inst type as base type for enum (details)
12. update the discussion of unit type (details)
13. Update 'fork+exec' test utility to be valid also in C and C++. (details)
14. Add EOF test coverage for string reading. (details)
15. Fix fork+exec missed case / broken build (details)
16. Parse string literals to get their length, taking into account escape sequences. Getting this from the lexer would avoid redundent work, but that is not set up. This corrects some typeof evaluations which lead to incorrect sizeof expressions. (details)
17. fix spelling (details)
18. Add uArray delayed-initialization equivalent to CFA array (details)
19. Hopefully fix broken build, from multiple declarations of arrah.hfa's delay_init. (details)

add void to lowerBound() and upperBound() declaration, which is a workaround to the warning message

update thesis

comment out unused variable declaration

fix dangling reference after delete

Made sure a variable is always initialized and removed some trailing whitespace.

Order of evaluation between arguments of a function is not left-to-right. Or right-to-left for that matter, it is unordred.

Removed SizeofExpr::expr and AlignofExpr::expr, expressions that would be stored there are wrapped in TypeofType and stored in the type field. Some special cases to hide the typeof in code generation were added. In addition, initializer length is calculated in more cases so that the full type of more arrays is known sooner. Other than that, most of the code changes were just stripping out the conditional code and checks no longer needed. Some tests had to be updated, because the typeof is not hidden in dumps and the resolver replaces known typeof expressions with the type. The __extension__ case caused some concern but it appears that just hides warnings in the expression which no longer exists.

Removed some unneeded code that could have gone last commit.

Add separate compilation discussion

Update thesis

Fix inst type as base type for enum

update the discussion of unit type

Update 'fork+exec' test utility to be valid also in C and C++.

Add EOF test coverage for string reading.

Fix fork+exec missed case / broken build

Parse string literals to get their length, taking into account escape sequences. Getting this from the lexer would avoid redundent work, but that is not set up. This corrects some typeof evaluations which lead to incorrect sizeof expressions.

fix spelling

Add uArray delayed-initialization equivalent to CFA array

Hopefully fix broken build, from multiple declarations of arrah.hfa's delay_init.

Summary

1. rephrase enum initialization (details)
2. remove spurious test files (details)
3. update test file covering problem cases (details)
4. add better error message for an empty enumeration, i.e., no enumerators. (details)
5. proofread intro chapter and add citation (details)
6. add citations (details)
7. move section from into to background (details)
8. formatting (details)
9. remove spurious push/pop from grammar contexts where type declarations are not allowed (details)
10. Implemented special case of sizeof/alignof, where reference types are ignored for these operations. (details)
11. add missing #ifdef to handle systems without io_uring (details)
12. modify time test to deal with unfreed storage from putenv/tzset (details)
13. preclude aggregate/enumeration type declaration in trait body (details)
14. After a years (or at least half a year) the CodeLocation optimization is merged in. Added Symbol (using Racket's name for interned strings), and used it for CodeLocation file names. The optimizes for the high number of copies - both instances with the same value and copy operations - and consistently brings down runtime by a few percent. (details)
15. update cast expression cost rule (details)
16. add void to lowerBound() and upperBound() declaration, which is a workaround to the warning message (details)
17. fix comment typo (details)
18. Fix #175 (details)
19. Improve inability to declare a struct member having a managed type via typeof. (details)
20. Add missing integration test of accordions. (details)
21. update accordion program, small changes (details)
22. added a few sentences and corrected spelling mistakes (details)
23. Added a check to make sure that optional parameters to not appear on function types not attached to function declarations. (details)
24. first attempt changing end-of-file to an exception (details)
25. Made heap pass the invariant check. We should be able to enable invariants on the standard library. Also added the reproductions for new trac trickets. (details)
26. update error message for parameter default-initialization in function-pointer declarations (details)
27. formatting (details)
28. Corrected handling of va_args in the parser. It still does not get through the resolver and further updates may be required. (details)
29. Removed two lingering set_extension functions from ast. There was one use each in the parser, which uses get/set functions still. (details)
30. simplify ttype min/max functions (details)
31. second attempt to simplify ttype min/max functions (details)
32. update citations (details)
33. remove empty line number increment from listings (details)
34. update thesis macros (details)
35. first proofread of chapter 3 (details)
36. first proofread of chapter 1 (details)
37. afterthought addition to chapter 3 (details)
38. Isolated the logic to decide if a function is a var-args function or not. (details)
39. citation updates (details)
40. proofread introduction chapter (details)
41. Changed the interpritation of () to be no parameters instead of any parameters. This had a lot of little changes because of this and some nearby clean-up. This includes some changes, including changing some generated functions to be fixed-args instead of variable-args, stripping out the place holder void parameter earlier, but it still shows up earlier in some cases that examine the parser directly. Also had to update the function generation tools. Have only tested with one --arch. Hopefully this all works out. (details)
42. correct check for EOF when reading character strings (details)
43. fix quotations from C++ reference manual (details)
44. Added checks for (and a test to check the checks) assertions we will not be able to adapt. Using an adapted version of Mike's error message. (details)
45. change exception throw to use copy constructor rather than assignment to initialize exception instance (details)
46. update thesis (details)
47. comment out unused variable declaration (details)
48. fix dangling reference after delete (details)
49. Made sure a variable is always initialized and removed some trailing whitespace. (details)
50. Order of evaluation between arguments of a function is not left-to-right. Or right-to-left for that matter, it is unordred. (details)
51. Removed SizeofExpr::expr and AlignofExpr::expr, expressions that would be stored there are wrapped in TypeofType and stored in the type field. Some special cases to hide the typeof in code generation were added. In addition, initializer length is calculated in more cases so that the full type of more arrays is known sooner. Other than that, most of the code changes were just stripping out the conditional code and checks no longer needed. Some tests had to be updated, because the typeof is not hidden in dumps and the resolver replaces known typeof expressions with the type. The __extension__ case caused some concern but it appears that just hides warnings in the expression which no longer exists. (details)
52. Removed some unneeded code that could have gone last commit. (details)
53. Add separate compilation discussion (details)
54. Update thesis (details)
55. Fix inst type as base type for enum (details)
56. update the discussion of unit type (details)
57. Update 'fork+exec' test utility to be valid also in C and C++. (details)
58. Add EOF test coverage for string reading. (details)
59. Fix fork+exec missed case / broken build (details)
60. Parse string literals to get their length, taking into account escape sequences. Getting this from the lexer would avoid redundent work, but that is not set up. This corrects some typeof evaluations which lead to incorrect sizeof expressions. (details)

rephrase enum initialization

remove spurious test files

update test file covering problem cases

add better error message for an empty enumeration, i.e., no enumerators.

proofread intro chapter and add citation

add citations

move section from into to background

formatting

remove spurious push/pop from grammar contexts where type declarations are not allowed

Implemented special case of sizeof/alignof, where reference types are ignored for these operations.

add missing #ifdef to handle systems without io_uring

modify time test to deal with unfreed storage from putenv/tzset

preclude aggregate/enumeration type declaration in trait body

After a years (or at least half a year) the CodeLocation optimization is merged in. Added Symbol (using Racket's name for interned strings), and used it for CodeLocation file names. The optimizes for the high number of copies - both instances with the same value and copy operations - and consistently brings down runtime by a few percent.

update cast expression cost rule

add void to lowerBound() and upperBound() declaration, which is a workaround to the warning message

fix comment typo

Fix #175

Improve inability to declare a struct member having a managed type via typeof.

Fix is evaluating typeof on both types in a cast; one was missing.

One layer of struct wrapping now works, with its test added.  Multiple layers of wrapping still does not work.

Add missing integration test of accordions.

Putting a CFA `array` inside structs was broken before the last fixes.

update accordion program, small changes

added a few sentences and corrected spelling mistakes

Added a check to make sure that optional parameters to not appear on function types not attached to function declarations.

first attempt changing end-of-file to an exception

Made heap pass the invariant check. We should be able to enable invariants on the standard library. Also added the reproductions for new trac trickets.

update error message for parameter default-initialization in function-pointer declarations

formatting

Corrected handling of va_args in the parser. It still does not get through the resolver and further updates may be required.

Removed two lingering set_extension functions from ast. There was one use each in the parser, which uses get/set functions still.

simplify ttype min/max functions

second attempt to simplify ttype min/max functions

update citations

remove empty line number increment from listings

update thesis macros

first proofread of chapter 3

first proofread of chapter 1

afterthought addition to chapter 3

Isolated the logic to decide if a function is a var-args function or not.

citation updates

proofread introduction chapter

Changed the interpritation of () to be no parameters instead of any parameters. This had a lot of little changes because of this and some nearby clean-up. This includes some changes, including changing some generated functions to be fixed-args instead of variable-args, stripping out the place holder void parameter earlier, but it still shows up earlier in some cases that examine the parser directly. Also had to update the function generation tools. Have only tested with one --arch. Hopefully this all works out.

correct check for EOF when reading character strings

fix quotations from C++ reference manual

Added checks for (and a test to check the checks) assertions we will not be able to adapt. Using an adapted version of Mike's error message.

change exception throw to use copy constructor rather than assignment to initialize exception instance

update thesis

comment out unused variable declaration

fix dangling reference after delete

Made sure a variable is always initialized and removed some trailing whitespace.

Order of evaluation between arguments of a function is not left-to-right. Or right-to-left for that matter, it is unordred.

Removed SizeofExpr::expr and AlignofExpr::expr, expressions that would be stored there are wrapped in TypeofType and stored in the type field. Some special cases to hide the typeof in code generation were added. In addition, initializer length is calculated in more cases so that the full type of more arrays is known sooner. Other than that, most of the code changes were just stripping out the conditional code and checks no longer needed. Some tests had to be updated, because the typeof is not hidden in dumps and the resolver replaces known typeof expressions with the type. The __extension__ case caused some concern but it appears that just hides warnings in the expression which no longer exists.

Removed some unneeded code that could have gone last commit.

Add separate compilation discussion

Update thesis

Fix inst type as base type for enum

update the discussion of unit type

Update 'fork+exec' test utility to be valid also in C and C++.

Add EOF test coverage for string reading.

Fix fork+exec missed case / broken build

Parse string literals to get their length, taking into account escape sequences. Getting this from the lexer would avoid redundent work, but that is not set up. This corrects some typeof evaluations which lead to incorrect sizeof expressions.