Webpage for development logs for GSoC 2015
Community Bonding Period
TODO list
- figure out if there is any major difference between .csg and .scad, except of expressions and modules
- figure out what parser to use (if it’s yacc/bison or re2c or any other, I’ll get involved and research what benefits does each offer)
- figure out how to use it, have a very simple demo set up within BRL-CAD build hooks
15 May Update
- made wiki pages for account, proposal, logs
- reread participation expectations and HACKING
- came up with the following grammar for parsing group() functions:
expr -> group_begin token group_end
token -> group_content token | expr
where group_begin is "group() {"
token is [a-z]+ "("*")";
and group_end is "}"
- looked over other BRL-CAD importer re2c lemon parsers ( wfobj and dom2dox) in order to understand how to integrate mine into BRL-CAD but with no luck so far
- currently investigating perplex tool and how is it used and integrated over the re2c layer
17 May
- dom2dox is not compiling with the rest of BRL-CAD's code so its CMakeFiles can not be used as an example of an integration of a re2c lemon parser
18 May
- cleared some empty holes in my understanding of the .csg OpenSCAD files format due to a thread on their mailing list regarding just that; it seems that there will be a new function added to the language soon, called trace() and also OpenSCAD2 will implement some changes in the existing methods
- got some help compiling dom2dox from Kalpit
24 May
- today was my birthday and I celebrated by erasing BRL-CAD and rebuilt it from svn sources to have a fresh start of Coding Period
Coding Period
Week 1
25 May : read about the lemon parser, tried to develop some simple numerical calculations in order to attain some skill using it
26 May : applied Cliff's csg makefile integration patch to my machine, built BRL-CAD successfully
27 May : small progress on the lemon tutorial , been occupied with school exam
28 May : finished the lemon tutorial, got some understanding of how lemon parsing is done, I have now a fully working calculator parser
29 May : not much work done today, posted an email on the mailing list with conclusions regarding the last few days work on lemon parser calculator
30 May : prepared for Monday school exam, no work done for GsoC today
31 May : got response from Cliff on mailinglist and cleared what I have to do next which is adding some simple parsing grammar for 1-2 csg functions (group() ,multimatrix() ) to the newly integrated parser in src/conv; started working on it
Week 2
1 June : worked on grammar, got some compiling errors, fixed a few, hope to have some valid parsing by tomorrow
2 June : more work on grammar, restrained the errors to just 1:
home/andrei/buildbrlcad/src/conv/csg/csg_scanner.c:1:0: error: ISO C forbids an empty translation unit [-Wpedantic]
I posted on mailinglist the error and the patch with the code that I wrote today.
3 June : prepared for next exam, did nothing for GSoC
4 June : further investigations on the empty translation unit error, not much progress , stayed on IRC
5 June : did not feel very well (headache and flu), did nothing for GSoC
6 June : got Cliff's mail response on my small grammar compiling error, applied the changes and it compiled well; segmentation fault at execution and gdb says the error is in PERPLEX_LEXER_private;
7 June : worked out the error and finally had group() parsed well
Week 3
8 June: just some code cleanup, prepared for tomorrow's exam
9 June: been at school for exam all day, did nothing for GSoC
10 June: tested and sent a valid csg parser patch for "group() {" ; started working on parsing "group(){ "some content" }"
11 June: worked on bonding group_content and group_end tokens with the existing grammar, almost done
12 June: the grammar that I proposed for group_content and group_end parsing is good but there seems to be a problem with the perplex scanner code; further investigations
13 June: prepared for Monday's exam, no work for GSoC
14 June: prepared for tomorrow's exam, no work for GSoC
Week 4
15 June: been at school for exam almost all day, did some small research on lemon grammar
16 June: started reworking on group() parsing, it seems that the latest patch didn't applied cleanly
17 June: sent the new patch which applies very well on my clean checkout; did some more work with the group_content parsing
18 June: further work on group_content parsing
19 June: finished the last exam, more work on group_content parsing, resolved some of the errors, hope to come up with a valid patch soon
20 June: finalised group_content and group_end parsing, made a patch, tested it on a clean checkout, posted it on sourceforge and sent a mail to the list with project updates
21 June: last patch got committed by Cliff, added tokens for all the functions/methods that can appear in a .csg file, started working on grammar for union() and difference()
Week 5 (pre-Midterm)
22 June: further work on the functions token, trying to expand the regex for group_content to cover all the functions formats in a .csg file and also trying to expand the grammar
23 June: made token for parsing dimension parameters for cube(), sphere(), cylinder() and polyhendron(), linked it with the existing grammar
24 June: designed non-terminals for assignation of parameters ( for example for parsing things like "convexity = 1" or "size = [3,4,5]")
25 June: heavy work on the csg grammar, lots of tests, working on edge cases, almost done & ready for a patch submission
Midterm Evaluation Period
At the beginning of the coding period I haven't had anything related to lemon and re2c integrated into BRL-CAD. I decided to take some time to study how to use this parsing tools before getting to the actual writing of the code. Because of my summer exam session, my progress was little in the first few weeks and, after working out some simple calculators using lemon, I finally started working on the .csg grammar. Initially, I though I could gradually parse and interpret functions and methods but I soon realised that having a fully functional grammar first is a better idea. With lots of help from Cliff and Andrei Constantin Popescu, I managed to finish the grammar just 2 days after the Midterm Evaluation Period opening and have it merged.
To run my code, just go in build/src/conv/csg , write an input file with some .csg text in it (or save a .csg there) and then just run ./csg input.txt output.txt . It should print on the screen various messages containing the tokens matched in the parsing process.
26 June: sent a patch regarding the .csg grammar which is almost complete , needs some work in the priorities section to be just right
27 June: filled midterm form on melange, worked on priorities
28 June: sorted out the priorities section, changed some of the internal structure of the grammar to parse imbricated functions better, added a new rule for assignation, made a new patch with the complete grammar and sent a mail to the list with the latest updates
29 June: talked with teepee from OpenSCAD about the conversion from .scad to .csg, found out how to properly test the proposed grammar (running the OpenSCAD test suite) and also got some insight about the functions that generate the .csg from the .scad ( toString for primitives and special functions )
30 June: did nothing for GSoC
1 July: tested my grammar using some .csg files provided by OpenSCAD, modified it to also parse "group();" , not only "group() {...}", added DEBUG macro and tried to print some matched functions using bu_vls_addr(&A->value) but with no luck
2 July: further work on printing matched functions, renamed some tokens in the grammar for a clearer code
Coding Period (2nd half)
3-6 July : had a mental breakdown because of the lack of progress and sleep in the last few days, took some time to organise my days and get back on track with everything
Week 6
7 July : restarted reading dom2dox code to learn about interpreting lemon grammar, understood some basic principles
8 July : cleared some confusions about the "END_TEXT;" macro after a discussion with Sean on IRC, decided not to use macros to prevent complexity issues
9 July : found the /doc/parsers/writing_perplex_lemon_parsers.doc and also the templates in there and read through them , they seem well explained and detailed
10-11-12 July : not much progress, tried a different approach by modifying the grammar to match functions as 1 token to reduce the number of terminals, ended up not parsing the function parameters
Week 7
did nothing for GSoC
Week 8
20 July : got Sean's mail about communicating more and staying more on IRC, tried to install tmux to be 24/7 on IRC but failed due to some incompatibility issues, decided to stick to XChat
21-23 July : documented about perplex, re2c and lemon in order to understand better what is not working and why and asking the right questions
24 July : sent an email to Sean and Isaac with questions about why fprintf(appData->outfile, "%s", bu_vls_addr(&A->value)) does not print anything but the token is matched, about the generated tree and also discussed updating deliverables
25 July : more work on sending matched strings to main, restrained the problem to a memory allocation issue