e-x-a.org » NPRG041 C++ (cs/en) 2019/20

Infopage about the C++ lab of NPRG041.

  • Taught by: Miroslav Kratochvil
  • contact: kratochvil@ksi.mff.cuni.cz (please include the course name/code in the subject, e.g. [C++])
  • Rule of thumb: When in any doubt, mail me.

Groups:

  • Czech group: every Wednesday at 10:40 in SU2
  • English group: every Thursday at 14:00 in SW2

Reading material:

Lab requirements

To get the credit, you have to attend the course reasonably (at least accordingly to your knowledge of the topic), do the homework (see below) and finish an individual project (also below).

You will be assigned several points for finishing the homework. These will be added to the points from the final test, therefore improving your final grade for the whole course. Details are available in the main course slides.

Depending on many factors, students from Erasmus programs may need a completely different set of rules to fit into different time limits. If you are an ERASMUS student, contact us!

Homework

Assignment A (deadline: end of November)

TBA

Assignment B (deadline: end of December)

TBA

Project

Each student has to create an individual project and submit it to get the credit. Topics of the projects may vary wildly, but you should discuss your topic before the end of November so that it is agreed upon. Size of the project is not an issue and largely depends on the topic, around 500 lines of (neat) C++ code is a good guideline (on the other hand, packing the same functionality into a smaller program using e.g. advanced language features is even better).

Bad topics:

  • Boring libraries
  • Boring games
  • Anything that does not behave C++-ish (e.g. code without a single class construction)
  • Anything that requires complicated or very specific user interface (e.g. bank account simulator, unless the interface is solved neatly in some novel C++ way)
  • Database workalikes, e.g. “evidence of hotel guests”, “evidence of soccer results”, etc.; unless the underlying database storage is somehow interesting.
  • Anything that has 100000 implementations already hanging around the internet
  • Also, there’s already too many of checkers, worms, tetris, pacman, snake, tictactoe, etc.

Good topics:

  • Small fast games (fun!)
  • Convenience programming libraries (convenient!)
  • Efficient data structures with demos (useful!)
  • Physics-like simulations of something (collisions, gravity, particles, etc. look cool)
  • Networking (internet!)
  • Virtual machines or interpreters for small programming languages.

Deadlines:

  • Topic agreed upon, written down in SIS: November 30th, 2019. Send me an e-mail to make sure the topic is confirmed and added to SIS.
  • Recommended time for submitting final version: March 2020.
  • Final version incl. documentation or example demonstration: April 10th, 2020. If you miss this deadline, we will assume that you are not aiming to obtain the credit. Projects that are first submitted after this deadline will not be evaluated at all. Corrections of previously submitted projects will be still possible.

Submission guidelines:

  • Make the code portable — it should not depend on the platform unless it is, by design, tied to that platform. Windows-specific projects should work on Windows in the computer lab, UNIX projects should work on Linuxes in the computer lab.
  • Do not over-engineer, avoid feature creep. The simplest project that satisfies the following conditions will do:
    • There’s a reasonable amount of C++ that shows you know what you’re doing
    • It does not crash, in particular it does not dereference invalid pointers, cause leaks, or torture the memory in any other way.
    • It does not contain any inefficiencies that could be fixed by better C++. (Repeat: const references! avoid manual allocation!)
    • It provably does what the topic says, and it can be demonstrated on an example use-case of reasonable complexity.
    • The code has sufficient comments. If you are unsure if you should add comment somewhere, try to tear the surrounding program block or function out of context, and ask yourself if anyone can still fully understand what it does (and why). If not, it needs comments.
    • If the topic is a data structure, include a comparison with a naive approach or some C++ counterpart (std::chrono provides timers). Note that your data structure does not need to “win” the comparison (that’s the topic of the HPC course). You should only provide a reasonable testing framework to assess the performance.
  • If you think the project is ready, either pack it up and mail it to me, or send me a git link.
  • Having received your program, I should be able to convince myself that it works in less than around 15 minutes. You can help it a lot:
    • Include a file INSTALL (.md) that describes how to make it work on a fitting computer configuration. (Better: How to make it work on computers in MFF’s computer lab. Best: Add a Makefile that works in the lab.)
    • Include a file DEMO or TUTORIAL that describes what should I do with the compiled program to see all the important functionality as quickly as possible. (Better: add a shell script that does it. Best: Support make demo)
    • If the DEMO requires some data for the demonstration, include it! (Advice: If I’m forced to create testing data by hand, it will take more than 15 minutes. Also, result will contain lots of hideous corner cases.)
    • If the source code is big, include INTERNALS file that describes which parts of the functionality can be found in which part/file of the code. This is sometimes also called “programmer documentation” or “hacking guide”. Imagine it as a signpost, compare with artist’s impression thereof.
    • If all documentation parts are favorably small, pack them together into one README.

Lab timeline

Source code from the labs will be available here.

Description and news from the lab will appear here.

Bonus material

  • Do not, never, ever use reinterpret_cast for converting pointers. Using a pointer converted by reinterpret_cast works by accident and may cause undefined behavior, EXCEPT in the case you reinterpret the pointer back EXACTLY to the original type (so you basically don’t do any conversion), provided the byte sizes of all values in the conversion chain are EQUAL. There are two legal usages: one for printing out a pointer address value as an integer, which you are usually not ever supposed to do; and one for converting binary data (e.g. creating floats from mantissa and exponent bits by hand), which you usually do not want to do. You want static_cast, or possibly dynamic_cast when downcasting base class pointers to derived class pointers. If you still want to use reinterpret_cast, remember there may be architectures where sizeof(char*) != sizeof(int*).
  • The Art Of UNIX Programming (online version) by E.S.Raymond is a brilliant material not only about the origins of current computing practice and the C programming language (that emerged together with UNIX and gave rise to C++), but also contains almost all the most important insights about software development gained in last 50 years. Surprisingly, all of them still apply. Sometimes a bit philosophical. If you are developing a large system (or a part of some) for the first time, this book will tell you what design mistakes you are almost certainly going to make, what are they called, and how to avoid them.
  • Doug Lea’s Malloc — a webpage about the original default allocator of most C/C++ libraries, which makes an extremely good read about allocators. (anyway — you probably use that guy’s code several times per second just by silently watching this website)
  • If you would like to create a game for the semestral project and have no idea where to start with the graphic output, try OpenGL. There are lots of ways to get an OpenGL viewport for your program, the easiest of them is probably the GLUT library. You might also want to see SDL that is traditionally used for portable games, or a newer alternative SFML. Following sites provide a good introduction to modern OpenGL: https://open.gl/ and https://learnopengl.com/.
  • If you would like your game to be a bit more advanced (and save yourself a lot of coding), use a graphics/game engine like Magnum (from a former MFF student), Ogre3D or Urho3D.
  • If you like template programming but the syntax seems unwieldy, use this :]

...