Changes

fix spelling mistake in error message

move unused routine log2_u32_32 from math to ready_queue library for later use, add integer ceillog2 as companions to integer log2, remove log2_u32_32 testing

update citations

adjust x86 testing after removing log2_u32_32

adjust arm64 testing after removing log2_u32_32

document no 128-bit builtin_clz for log2, remove bogus 128-bit test for log2, add companion ceillog2 tests

adjust x86 testing after updating integer log2/ceillog2 tests

adjust arm64 testing after updating integer log2/ceillog2 tests

do content note from Pat: use a concrete acrostic poem

do content note from Pat: justify performant string for C

add acknowledgements

de-dangle references, kill orphans, and otherwise run amok on superficials

do last typos from Pat

allow making print version without editing source

add usage comment to string perf tool for ad-hoc runs

Fix bug where string append leaves heap corrupt for compaction.

string_res.cfa/function "append": do the fix
string_res.*/"DEBUG...": enable test to check that compation occurred
tests/*: add test of the bug case; this test fails without the fix

Tested for no impact on string concatenation performance.  For repeated
concatentations, the hot path is the "already at end of heap" case.
For concatenations spread across many strings, copying them to the end
of the heap dominates.

add more comments to hasIntegralType fix, add tests for references in switch/choose expression

normalize compiler-flag predicate names with pass names used with -P command-line flag

document and test found math function log2_u32_32

copy edits from Pat

allow reference typed expressions in switch statement

add comments related to loop-else clause

copy-edits found during a pass-through

clarify multiple link axes undesirable pattern

weaken conclusion sentence to circumvent an objection from Pat

reduce figure 2.9 to within the text boundary

alvin thesis scratch ideas

copy edits from Patrick

copy edits from Gregor; content change requests from his 'scribble' noted as 'MLB'

address Gregor's 'scribble'/mlb content remarks in Background.

add .s results to bound check add-remove experiment and run with arm and clang

revise array bound-check removal experiment clang version to use LLVM's libc++

address Gregor's 'scribble'/mlb content remarks in Array

address Gregor's 'scribble'/mlb content remarks in List

flesh out multi-byte character discussion

remove inaccurate unnecessary character-set info from early string into

add remarks about microbenchmark potentially failing to generalize

add expplicit what-is-a-string qua struct

elaborate string char* compatibility discussion

clarify my rejection of most warnings as type errors being due to unsound detection

address Pat coments on C per standard and presentism of C evaluation

several content edits from Pat

reorganize the list api coverage into one narrative without cross-referencing a separate figure

visual updates to thesis material

remove utf-8 characters from bibliography

Updated the exception proposal. Major additions are sketches of the implementation and panic system.

fix Mesa citation

create PhD thesis directory for Fangren and populate with basic latex macros and emails

update PAB exception proposal

update uC++ to CFA manual

update lexing and documentation for new constant separators

rename paper

small changes to constant separator section

update comment

add digit-separator characters ' ' and '\'', along with existing separator '_'

formatting

formatting, ignore stack_verify if preemption occurs during coroutine context-switch, rewrite preemption_ready routine

change meaning of < for user type

complete polymorphic abs defintion, change all max functions to use !<

formatting

add user type to abs test

formatting

change max from trait > to < and use !< for >

add OFSTREAM_VOID and OFSTREAM_VOID_IMPL for ofstream to match with OTYPE

update OSTYPE_VOID_IMPL macro to use typeof

update tests to use OFSTREAM_VOID_IMPL for void output

update citations

add papers on modules

add groovy declaration for NewAST

try removing NewAST

add citations and papers to Alvin's thesis

remove build directory from repository

back out of adding more compilers

remove blank lines from groovy script

try more gcc compilers for full build

try the full build with gcc-12

formatting, turn off -Wdangling-pointer around small code fragments because of false positive

formatting

turn on -Wdangling-pointer conditionally for gcc >= 12

change unnamed enumeration to named and hide the enums to prevent name clashes

fix subscript error in expression.cfa

comments

formatting, add -fwrapv compilation flag to fully define integer signed overflow

restructure test program to eliminate warning message

remove volatile workaround for the missing -fwrapv compilation flag

move libcfa/src/concurrency/atomic.hfa to libcfa/src/bits/atomic.hfa, and update all the atomic operations

change routine name from rdtscl to rdtsc

change function name isEmpty to empty

fix problem with printing 0 for largest int_128 values

formatting

change routine name from rdtscl to rdtsc

fix -O2 problem with new atomic macros

elide warning message -Wdangling-pointer

fix strange -O3 gcc bug by adding volatile

make ignored "-Wdangling-pointer" conditional on gcc >= 12

fold temporary for const_cast into single expression

revert test gcc-12 in full build for x86 and x64

formatting

remove addDataSectionAttribute hack and handle dynamic initialization of const definitions by removing the const after resolution

formatting

remove enum for PTHREAD_KEYS_MAX and replace with #include <limits.h>

formatting

test gcc-12 in full build for x86 and x64

Added to the Assertion Like Exceptions proposal, just a little section to address checked exception concerns.

Rewrote part of the assertion like exceptions proposal, combining Syntax and Usage section with the Examples into Features.

update Default and Named Parameter section

differentiate between C _Alignof and gcc __alignof__

fix spelling mistake in directory name

add thesis template for Alvin's thesis

temporary patch parsing attribute location in function prototype

Six months to a year after it was due for a round of early feedback: The original exception rework proposal.

fix loops incorrectly accessing header node, change from serr to printf for debug

fix sense of assert condition

updated exception proposal from PAB

Prototype code for modules

yet another exception proposal

remove string 'contains' function and replace with 'includes', make string 'includes' inline

formatting

harmonize output quote manipulator with input quote manipulator, adding left/right quote parameters

update printing of decimal point in floating-point numbers

fixed x86 expect file for manipulatorsOutput1.x86.txt

formatting, add some tostr conversion routines

add quote output manipulator for printing char, char *, string surrounded by quote characters

formatting

replace utf8 single quote with ascii single quote

update user documentation with new quote manipulator

Slightly loosen a test that has failed on 32-bit build lately.

Since 16d9c3a, this test was failing on all builds.  The present change in not 32-bit specific; it's expected to fix this particular test, on all builds.

Note there are further 32-bit issues remaining, even after this fix.

Test is:  array-collections/c-dependent

Loosening is:  Adjust test's scope to avoid exercising "truly incorrect" C code; which may receive compile-time array-bound warnings on sufficiently advanced C compilers.  In test code and .expect, switch from passing incorrect bounds to passing correct bounds; the point of the test (newly clarified, always foremost) is that CFA handles C's dependently sized VLA parameter.

formatting

formatting

Strengthen a test that had been evading warnings on 64-bit only.

This test change is expected to break the 64-bit build (to match currently-broken 32-bit).  A subsequet test change is expected to fix both builds.

This change makes a test, which was incorrectly architecture-specific, into one that is architecture-independent.

Test is: array-collections/c_dependent.

The relevant array-size warning occurs in gcc-11+ (new-gcc).

The test had been passing on 64-bit only because of a quirk: new-gcc's dependent warning gives false accept when a dimension expression is complex enough; CFA's (arch-specific) inserted casts interacted with an incidental choice of the test program, as follows:

// original
void f__bound_ptr_allow( int n, float a[n] ); // source
void f__bound_ptr_allow( int n, float a[(unsigned long int) n] ); // 64-bit lowering => different enough type, check skipped
void f__bound_ptr_allow( int n, float a[(unsigned int) n] ); // 32-bit lowering => close enough type, check enforced

// this change's revision
void f__bound_ptr_allow( size_t n, float a[n] ); // source
void f__bound_ptr_allow( unsigned long int n, float a[(unsigned long int) n] ); // 64-bit lowering => close enough type, check enforced
void f__bound_ptr_allow( unsigned int n, float a[(unsigned int) n] ); // 32-bit lowering => close enough type, check enforced

2nd attmept to build x86 on brokk

1st attmept to build x86 on brokk

assure consistent section title capitals

establish first-person singular use upfront

change to consistent "collection" over "container"

make table-caption capitalization and length match figures

revise list perf histogram introduction

fix mistakes in use case numbering in list perf results

drop unnecessary premature sentence

take pg1 out of the weeds on c--rust translation

add more array future work

tidy list 1ord discussion with refreshed numbers and stronger commentary

spelling corrections, and final proofreading

Apply document-level formatting and rearrange the array-in-struct discussion.

Formatting:
- add visual separation of figures from text
- prevent "formula"-style code fragments from page breaking
- (thereby prevent code formula broken at end of page from mistakenly restarting the reader in a code figure at the top of the next page)
- give short+long figure captions
- assure descriptive captions
- remove Title-Like Captialization in Captions
- turn a couple figures into tables where the figure visual separator looks bad

Array-in-struct
- remove lingering use of "accordion"
- call it Dynamic Array Member
- oranize discussion as, first explain example, then argue feature is good

proof-read array chapter

temporarily off-line Jenkins x86 builds

fix error in Jenkinsfile

proofread list chapter

proofreading in string chapter

proof-read through background chapter

further background proofreading from "forgot to press save" before last commit

fix thesis broken build from incomplete deletion of a graph

address two string-section oustanding items: remove confusing sentences; add attribution plot overlay for separate compilation overhead

fix list perf plotting issues with physical factors

address string pending item, to elaborate on the cfa "text import" slowdown

Revert 75ba2fa6b formatting on plots.  Aligned plots use aligned axes.  All plots have log vertical.

revisions to ll perf intro and graph formatting

staged modules proposal (updated)

staged modules proposal

Data supplement for list perf auxiliary analyses

Anticipated last new content for list perf analysis

save draft of list perf fx-interaction plot

Add overlaid means to list perf histograms.  Add 2nd-order graph to the paper and its discussion.

Retract work in progress mistakenly included in 1abcec9b.

establish and nail down definitions of operations and size zones

changes to first-order effects

working on Section 4.6.3

refactor list perf analysis with histogram view reusable

adjust list plot analysis making drilling independent from marginalizing/conditioning

fix physical-effect histograms (by providing right arguments on above knobs) to be completely unconditional (which they always claimed to be).  difference is noticeable but not story-affecting.

Add the list perf non-physical comparison (as promised).

This commit is a save-snapshot before I pivot this comparison to compare all first-order effects, which really is the right context.

switch list perf winner-loser plot to uniform treatment of all first-order effects; add discussion of key conclusion that fx is not your big-ticket item

proofreading list chapter

Refactor list plotting code to enable new analyses.

Visible difference in relative plots no longer being 3 others baselined to lq-tailq, now all 4 baselined to their mean.

Revise presentation of absolute times and add discussion of their physical influences.

small updates to list chapter

update experimental machine descriptions

update and add citations

additional updates in string chapter

proofreading in front half of list chapter

3rd attempt at harmonizing isOp functions, e.g., isListed, isFirst, isLast

2nd attempt at harmonizing isOp functions, e.g., isEmpty, to C/C++ form empty

temporary fix for auto-gen destructors using linkonce

1st attempt at harmonizing isOp functions, e.g., isEmpty, to C/C++ form empty

fix signal_block on empty condition queue

add tuple-type insert and remove functions to list type

add one more item to future work in conclusion

move latex savebox definitions, add package multicol

use subfloat for figure programs, inline footnote

final proofread of conclusion

remove attempt to make C inline comments look better in latex output

harmonize AST dump and flag names, and the order the passes are called

update user documentation on AST dump flags, and turn off silent in Makefile

rename predicate flag that conflicts with AST dump flag

final proofread of background chapter

final pass of abstract

final proofread of introduction chapter

rework member fixBlock to allow loop else-clause to access while/for conditional declarations

change codegen for loop else-clause to print else-clause as a compound statement

during parsing, wrap loop else-clause in compound statement if it is only an expression

update loop else-clause test to verify loop declarations are accessible in else clause

update documentation for hoisting of loop declarations

2nd attempt to turn on ARM builds on algol

adjust name for ARM architecture

formatting

correct spelling from valgind to valgrind

update choose and waituntil statements

turn on ARM builds on algol

add routines stack_verify and stack_pointer, in debug mode call stack_verify on front-side of context switch and time-slicing

increase user-thread stack from 64K to 256K

remove making keyword exception an identifier for math.h

formatting

add PLG-specific instructions for running antlr

addressing comments of module proposal

first proofread of module proposal

update citations

list perf "zoomout" numbers run with interleaving and observation disabled, which removes an anomaly and reproduces the "zoomin" numbers in the "sweet spot"

LL perf: adjust configurations

Simplify doing a manual one-off run.  Disable harness's iterators on "zoomout" plots.  Remove cfa-strip and add lq-list for "zoomout" plots.  Add more exploratory analyses.

LL perf: small fixes

Fix off-by-one bug in runtime-no-shuf solution.  Add safety check that interleaving is off.  Re-run.

Replace dlist test's sout with printf, to enable testing non-libcfa draft implementations

Move link-list to-be implementation to libcfa

Make dlist2 pass the original dlist test.  Now, there is one common test for both.

Data and analysis associated with harness of commit 78bc398830.

Analysis change narrows the sweet-spot zone to where CFA is not misbehaving on queues and where general noise is reduced.

Revised plots are going into thesis doc, but commentary/focus has not shifted accordingly.  Notably, the CFA-attribution analysis is now mostly obviated.

Various changes motivated by improving CFA score on len-1 queues.

No such CFA score improvement achieved.  Each change helped only on stripped-down, "try to isolate an important factor" tests.  Generally, the changes are benign refactorings.  (Results substantiating "don't hurt" are forthcoming.)

Libcfa changes are
- move a read action from between the memory breaks to before them
- make the memory breaks conditionally excluded (default included, as before)

Harness changes are
- add width, a compiled-in number of lists to use in round-robin order; defaults to 1, which is what was happening all along
- make the analysis scripts tolerate (so far, ignore) the width column
- rename CLI arg NumNodes to Length (now NumNodes is Length * Width)
- factor core testing loops into helper function `runtest`
- switch to signal-based termination (and add uC++ work-around)
- put "iterator threading" into ITERS_SAVE, joining preexisting "save iter into elem's self ref"; make iterator threading conditional on iterators-active
- disable insertion loop counter and obs_*-variable declarations (and thus writes) when observation disabled
- generalize observation to work on multiple lists
- make observation and assertion-check-enabled mode work on stripped CFA list implementations like tagging-disabled
- through this observation, ensure correctness of multi-list versions

Adjust stripped-harness comparion config to exclude interleaving, matching current general runs

Add data for Intel host 'java', alongside incumbent AMD host 'swift'.

Revise analysis--presentation to show both side by side.  Remove presentation of CFA Attribution, which is now seen to be chasing a red herring.

Data continue to be from harness of commit 78bc398830.

Refresh data to harness of 9d3dc40802c6.  Latest harness and data now correspond.  Plots in built thesis show latest data.

move old module work into another folder

stitched modules proposal

2nd attempt to turn off ARM builds on algol

try to turn off ARM builds on algol

for opening a file, change to default initialization of mode parameter from overloaded functions; fix fstream initialization bug by calling default constructor in other constructors

formatting

fix wording issues in last proofread of array chapter

last proofread array chapter

Revise array's overload-selection comment now that #247 is closed.

The set of overloads required is unaffected, just there's a simpler derivation of this set.

Improve libcfa-array's bound-check removal and write that thesis section.

The libcfa change adds a more performant alternative for a subset of multidimensional indexing cases that were already functionally correct.
That the new alternative is more performant is not shown in the test suite.
There is an associated new high-performance option for passing an array-or-slice to a function.
The added test cases cover those options.

The added in-thesis demos rely on the new more-performant alternative for multidimensional indexing.

proofread string chapter

small wording changes

more array proofreading

Edits in early sections.

Includes explicating the connection between unsafety stats and the three collection types.

proofread background chapter

proofread introduction chapter

proofread abstract

rename private waituntil routines register_select, unregister_select, and on_selected with tailing $

harmonize single_future with other future types, remove multi_future, marks its test as deprecated, and turn off its test

more cleanup of future code

fix design error in test waituntil/futures.cfa

add reference counting futures for use with waituntil

many updates and additions to the uC++ to CFA cheat sheet

raise * in /* and */ to look better

many citation updates

add missing Ada test program to Fangren's thesis

update semaphore lock

rewrite test program with futures and waituntil

Add issues with autogen removal: hiding too eagerly and not acting inspectably

change name is_empty to empty for monitor conditions

change use of random to prng

update semaphore lock

formatting

new version of barrier lock

add shadow variable to thread structure

change counters from ssize_t to size_t

rename private variable with trailing $

rename private variable with trailing $, and restructure lock

add "clear" function to strstring

update barrier tests for updated barrier

remove warnings from empty verify macros

Indicate CFA only builds on Linux operating system

add bibliography references

Add section about contributing to the Cforall project

test full build on x86

2nd attempt test full build on ARM

update anew/adelete ctor/dtor to handle array with 0-byte elements

formatting, add tests for new anew/adelete

update expected alloc-ERR files

test full build on ARM

remove empty chapter benchmarks.tex

wording and formatting changes to multiple sections

fix symbolic links from concurrent to concurrency in longrun_tests

update strings, update for-control and string documentation

harmonize user manual string discussion with string chapter

make tabcolsep variable global

clean up string discussion