- use explicit expression lists instead of binary trees to represent lists of the form a, b, c
(per discussion w/ Russ)
- use explicit nodes for various language constructs for better readability
- various adjustments in parsing and printing
next steps:
- clean up AST fully so it can be checked in as library
R=r
OCL=26371
CL=26371
- various parser fixes to match updated spec (&&, &^=, label decls, const decls)
- using html template for directory and error page in doc server
- show compile errors inplace in the source
- cleanups
R=rsc
OCL=26287
CL=26287
- correctly associate comments with declarations
(available through AST)
- very raw printing of interface
- much more functionality, now needs some formatting, sorting, etc.
R=r
OCL=26213
CL=26213
if T has pointer methods. this is just a heuristic
but it catches the problem robert ran into and
lets me put the larger interface issues aside for
now. found one bug in pretty.
R=ken
OCL=26141
CL=26141
- removed any symbol table stuff
- some interface cleanup
- removed lot's of dead wood
- half the support for type switches (simple form)
- name capitalization
Several rounds to come.
R=r
OCL=25797
CL=25797
- move scanner to into lib/lang
- added test
- adjusted various make and build files
R=r
DELTA=1731 (973 added, 753 deleted, 5 changed)
OCL=25668
CL=25713
- separated out token definition from scanner
- cleaned up token and scanner implementation
- added comments to exported interfaces
R=r
OCL=25665
CL=25665
- require ()'s around composite literals at the if/for/switch control clause level
- fixed a nasty bug: passing a value instead of a pointer to a value to an interface
variable - and not noticing that the value is copied
R=r
OCL=25649
CL=25649
- removed unneeded code that accumulated over time
- change src from string to []byte (perhaps should be io.Read
but that has some other disadvantages)
- simplified interface
R=r
OCL=25615
CL=25615
- converted more of AST and parser to use interfaces and explicit
structs for individual Go constructs (can be replaced now with
interface calls such that the parser becomes AST structure
independent, as suggested by rsc)
- added more tests (find all .go files under GOROOT)
- (temporarily) lost html links for identifiers when generating
html output
- TODO: lots of cleanups
R=r
OCL=25518
CL=25518
- moved Object, Type, Scope out of AST into symboltable
- moved universe into symboltable
- removed dead code
- fixed dependency computation (pretty -d filename.go)
- lots of cleanups
- removed tocken channel connection between parser and scanner
(was cute, but not really needed)
R=r
OCL=24545
CL=24545
using interfaces properly => much cleaner code
- converted tracing code to use 'defer' statement
- next steps: convert rest of ast as well
R=r
OCL=24277
CL=24277
- fine-tuning of ast
- more accurate block pos info (improved printing in some cases)
- collecting local variables and fields
- more work on type checking
- lots of minor tweaks
R=r
OCL=23375
CL=23375
- consolidation of files, cleanup
- more success producing idempotent output for some files with comments
containing tabs
- snapshot of the day
R=r
OCL=22474
CL=22474
(this code doesn't match the existing language at this point,
but it's a large code base which compiles - will eventually go
away)
- enable compilation of it again in run.bash
R=r
DELTA=1147 (534 added, 311 deleted, 302 changed)
OCL=22176
CL=22176
(will help produce html output where we can click on identifiers and
get to the declaration)
- snapshot before changing back to old new
R=r
OCL=22159
CL=22159
Preparations to track identifiers and declarations so that we can
generate good html links as pretty printer output:
- brought over old code and adjusted it
- initial hookups, nothing really running yet
R=r
OCL=21383
CL=21383
- fixed html tag generation
- simplified html escaping machinery
(not 100% correct for strings yet)
- can now produce the first mostly correct formatted html pages from source
with (fake) links: e.g. pretty -html srcfile.go > srcfile.html
R=r
OCL=20915
CL=20915
mark and sweep, stop the world garbage collector
(intermediate step in the way to ref counting).
can run pretty with an explicit gc after each file.
R=r
DELTA=502 (346 added, 143 deleted, 13 changed)
OCL=20630
CL=20635
can run peano 10 in 100 MB (instead of 1+ GB) of memory
when linking against this.
can run peano 11 in 1 GB of memory now.
R=r
DELTA=100 (44 added, 44 deleted, 12 changed)
OCL=20504
CL=20553
- first cut at html writer (will do html escaping, html tag production)
- first cut at generating basic html output via pretty
- some cleanups
R=r
OCL=20550
CL=20550
- by default consider extra newlines in src for better formatting
- additional flags for control (-newlines, -maxnewlines, -optsemicolons)
- don't print ()'s around single anonymous result types
Status: Comparing the output of pretty with the input for larger files
shows mostly whitespace/formatting differences, which is what is desired.
TODO:
- Handling of overlong lines
- some esoteric cases which look funny
R=r
OCL=20293
CL=20293
- changed tab width to 8 chars by default to make test run properly
- excluded 4 files that are not idempotently printed
- fixed a couple of incorrect file position recordings in parser
- fine-tuned layout engine
- white-space formatting reasonable, but not perfect
- no handling of overlong lines
R=r
To use with smaller tabs: pretty -tabwidth=4 file.go
To use with blanks instead of tabs: pretty -usetabs=0 file.go
OCL=20184
CL=20184
- comment printing still disabled by default because idempotency test fails
- whitespace control better but not perfect yet
- snapshot before making some heuristics changes
R=r
OCL=20151
CL=20151
for end of declarations, blocks, parameter lists, etc.
- use extra src positions to more accurately print comments
- fine-tuned low-level printing routine for comments
- added better debugging support
Status:
- comments now appear at the right place (inbetween the right tokens)
- newline control needs improvement (not very hard)
- comment printing disabled for now because pretty is not idempotent
with it; to enable: -comments
R=r
OCL=20079
CL=20079
(to be moved into std lib eventually)
- rewrote tabwriter to use byte buffers instead of strings
(byte buffers to be moved into stdlib eventually)
- support for recent syntax changes
- no space printed after function name and before function parameters
- comments still disabled due to a known bug
R=r
OCL=19430
CL=19430
- implemented elastic tabstops algorithm, now correct and documented
- first cut at printing comments (use -comments flag, disabled for now)
- struct field types are now aligned (using elastic tab stops)
- needs more fine-tuning
* fixed a bug in test script
* added quick smoke test to makefile and invoke it in run.bash
instead of the full test
R=r
OCL=19220
CL=19220
- new code enabled, but no comments printed yet (so the effect
of the elastic tabs is not seen yet)
TBR=r
DELTA=200 (93 added, 69 deleted, 38 changed)
OCL=18951
CL=18951
