All compiler programming projects must meet these requirements on a standard student account of a CTB60 ("alamode") machine.
Requirements
As a placeholder for this wiki page, we'll use PROGRAM since it reads nicely in prose. For compiler projects though, the name will be something else.
Submitted as a .zip or .tar.(Z|gz|bz2|xz) compressed archive. Do not use RAR archives.
If you are generating an archive with symbolic links from another *nix box, just use tar --- the invocation of zip seems to be tricky (or there are inconsistent zip versions in Linux land).
The unrolled archive will be scanned for a top-most key file: only one of these may exist:
A Makefile compatible with GNU make (this is the standard make on a Linux box)
A Build.sh Bourne (sh) or bash script
Case in names are important! It is Makefile with a capital M, not makefile; likewise for Build.sh. This is a 400 level course in a CS program, I think it is reasonable to expect students to be able to either
- develop in a Unix environment
transport a system independent code base1 from Windows or MacOS X to a Unix environment.
If you need help doing this please let me know, I'm happy to assist.
In either case, after running2
top $ make
ortop $ bash Build.sh
there should be an application3 named PROGRAM. 4 Note that PROGRAM is always in all capital letters.
Your submission must build without extra installs of additional packages or libraries to the grader's account. The only exceptions to this requirement are the Python graphvis package and some select student favored languages: rust, go, lua, .NET. See this page for details about using one of these languages.
PROGRAM will be invoked with the appropriate command line arguments for the assignment, results required by the assignment should be printed on stdout.
Any stdout output line with a required result must have OUTPUT as the first word (token).
- All numerical values required by an assignment must have whitespace (which includes end of line) around them.
Required textual output should be surrounded by single colons (:).
The order for results is assignment specific and important, they must emanate from the application in the order dictated by the assignment.
Depending on the assignment, PROGRAM may be required to write a specifically formatted result file, in which case the OUTPUT rules above won't apply to the contents of the file.
Your submitted archive should not contain unnecessary files (.o, .pyc, *.h~, Cargo.lock ) or executable machine or vm code (*.class files, a __pycache__ directory, bin, target or obj directory).
Files and directories that begin with an underscore are reserved for use by your instructor, TA, or grader.
- Your submitted archive file should not have any entries that begin with an underscore.
Your PROGRAM should not use or create files or directories that begin with an underscore.
Written answers or graphs should always be provided in a doc subdirectory, the doc directory should be alongside your Makefile or Build.sh.
- Prose should be provided in either a text file, an HTML file, or a PDF file.
- Graphs may be embedded in your written answers (if you provide HTML or PDF). If your answers are written in a simple text file, provide your graphs in a common raster image file format (PNG, JPEG, ...), SVG, or (preferably) PDF. Refer to the graphs by name in your answer write-up.
Graphs provided for assignments without a required written component should be in appropriately named files (still, of course in doc) and should be self-explanatory on viewing.5
Students may always feel free to submit a README file for any assignment, but it needs to be in doc and follow these conventions.
Your PROGRAM should not prompt the user for any input, nor pause the display of console messages for the human eye.
Examples
Submission Templates
Here are example trees for submissions, one each for the most common languages used by students. If you are using another language, look at one of the templates for a similar language.
Templates with a Makefile and a Project.mak: edit Project.mak which is the intended way to configure the build.
Rust language using build script - contact instructor if you have issues.
.NET language using build script - contact instructor if you have issues
Here is a source tarball that is a "more pure" example of a non-trivial Makefile for building C++. This example doesn't use my Project.mak build configuration file.
If you like to use IDEs such as Eclipse or M$Vs:
- do your work in the IDE with a simple source layout
- copy your code into the appropriate extracted template on a CSM Linux box
test to make sure it builds and runs correctly
clean it out (there is a clean target in the provided Makefiles, otherwise remove residual files by hand)
tar or zip it up and submit.
If your IDE purports to generate Makefiles, then give it a try (YMMV). Be sure to test on a Mines machine!
Beware that M$ based IDEs frequently generate NMAKE files, which are a far cry from conventional Unix make compatible build files, and are uniformly mocked by gmake.
Output Formatting
Suppose a compiler assignment dictates:
Your PROGRAM should accept two arguments: the first is a file containing a grammar, the second is a valid sentence of terminals from Σ. PROGRAM should emit the following values:
- The total number of productions performed by the grammar.
- The maximum size of the parser engine stack during the parse.
- The type and number of each non-terminal node in the parse tree.
Then both of the following outputs would be acceptable (they differ only in the amount of programmer debug messages spewed):
$ ./PROGRAM grammar.cfg "int a=3, b=4; a+b<6 if b, a = a, b; else { a=b=0; } output("else");" OUTPUT 63 54 :DECL: 1 :EXPR: 6 :OUTPUT: 1 :STR: 1 :PROG: 1 :IFELSE: 1 :EQN: 5 $$ ./PROGRAM grammar.cfg "int a=3, b=4; a+b<6 if b, a = a, b; else { a=b=0; } output("else");" Read grammar 'grammar.cfg' 6 nonterm 17 terminals EQN production a 3 EQN production b 4 EQN production b a EQN production a b EQN production b 0 EQN production a b STR production else OUTPUT counting non-leaf nodes... 63 OUTPUT 54 OUTPUT nonterm/0/7 :DECL: 1 OUTPUT nonterm/1/7 :EXPR: 6 OUTPUT nonterm/2/7 :OUTPUT: 1 OUTPUT nonterm/3/7 :STR: 1 OUTPUT nonterm/4/7 :PROG: 1 OUTPUT nonterm/5/7 :IFELSE: 1 OUTPUT nonterm/6/7 :EQN: 5 fini $
But this one is not acceptable for two different (unique) reasons, do you see why?
$./PROGRAM grammar.cfg "int a=3, b=4; a+b<6 if b, a = a, b; else { a=b=0; } output("else");" Read grammar 'grammar.cfg' 6 nonterm 17 terminals DECL production int EQN 1 ;s EQN production a 3 EQN production b 4 DECL production int EQN EXPR production < EXPR production + EQN production b a EQN production a b 2 ;s EQN production b 0 EQN production a b IF production EXPR EQN EQN 3 ;s STR production else OUTPUT production STR 4 ;s Parse tree complete 63 nt nodes 38 term leaves OUTPUT 63 OUTPUT 54 OUTPUT nonterm 0 of 7 :DECL: 1 OUTPUT nonterm 1 of 7 :EXPR: 6 OUTPUT nonterm 2 of 7 :OUTPUT: 1 OUTPUT nonterm 3 of 7 :STR: 1 OUTPUT nonterm 4 of 7 :PROG: 1 OUTPUT nonterm 5 of 7 :IFELSE: 1 OUTPUT nonterm 6 of 7 :EQN: 5 fini $
The applications in this course should be pretty OS agnostic (1)
The name of your directory containing Makefile or Build.sh does not have to be named top, the example is simply showing that make and bash will be run from within your top directory. (2)
This means the Unix executable bit is set. (3)
This does not preclude PROGRAM existing before make or Build.sh are run, and a build script might be a mere placeholder. Such a setup is applicable for certain languages. (4)
Heaven forbid, don't forget a graph title. (5)