|
GCC reports different versions for different platforms, meaning I can't compare the version of GCC i'm running on windows with the one that's running at Travis-CI.
The trouble is my code compiles fine locally, but fails at travis-CI and compiler version differences are the only things I can think of at this point that would explain it.
I was hoping to optimize my C++14 rendition of my GFX code a little, and bring it up to par with the C++17 version but the g++ compiler at travis-CI will not eat my code no matter how I coax it.
I'm completely stumped and frustrated because my morning started on an optimistic note - i saw an opportunity for optimization - and I was rewarded with broken build after broken build. 
EDIT: AHA Was told how to update the distro running at travis-CI and it worked.
To err is human. Fortune favors the monsters.
modified 13-Apr-22 14:05pm.
|
|
|
|
|
C++11 FTW. 
|
|
|
|
|
The problem with that is C++11's ability to evaluate at compile time is extremely limited compared to later versions of C++, and I routinely open a dialogue with the compiler in my code. Basically, my code relies heavily on compile time evaluation of complicated expressions - even bit shifts that aren't byte aligned. You just can't do it with C++11 or I would have. C++14 is the lowest one I can target to get the features that I need.
To do what I'm doing in C++11 is simply not possible. My code could not function the way that it does now. There's simply not language support for it.
To err is human. Fortune favors the monsters.
|
|
|
|
|
I'm glad to see that you updated your original post to say you got this sorted. Now I don't feel so guilty for trolling. 
|
|
|
|
|
You need to google the "Works on My Machine" sticker.
The difficult we do right away...
...the impossible takes slightly longer.
|
|
|
|
|
GCC is a ![Elephant | [mastadon]](https://www.codeproject.com/script/Forums/Images/elephant.gif) joke. Easily the worst compiler I had the displeasure of using, and I compiled VB6 software. The fact that linker arguments have to be passed in a specific order if a library calls some other library is the most bonkers "feature" I ever saw, and the error messages passed by the compiler are gibberish when compared to... well any other compiler out there (MSVC, Intel, GreenHills, Keil to name a few).
GCS/GE d--(d) s-/+ a C+++ U+++ P-- L+@ E-- W+++ N+ o+ K- w+++ O? M-- V? PS+ PE Y+ PGP t+ 5? X R+++ tv-- b+(+++) DI+++ D++ G e++ h--- r+++ y+++* Weapons extension: ma- k++ F+2 X
|
|
|
|
|
GCC is a fine compiler. It also happens to have a pluggable backend meaning it's a much better cross compiler than most other offerings
In my opinion MSVC is garbage I don't even target it with my code, and which is probably why VS supports clang now.
But of course we'll never agree on this.
To err is human. Fortune favors the monsters.
|
|
|
|
|
|
UML diagrams are useful for documenting an architecture--how classes collaborate. Beyond that, I look for good interface and implementation comments. Markup like Doxygen is noise that gives me gas. Comments should be in proper sentences that describe what code is going to do (interface comments) and how it's going to do it (the main focus of implementation comments). Comments that describe a few lines at a time are often annoying, no different than people who constantly interrupt. It's usually better to comment larger blocks of code.
|
|
|
|
|
I write technical articles on using my code, with code examples. On top of that, I often generate reference docs using a tool like doxygen.
I find that creating a technical article not only makes it easier for people to get their feet under them using it (at least I think so), it also helps me later to flesh out an outline of the major components of the software, which then evolves into a table contents. Finally that becomes master documentation.
Tools like Doxygen can only supplement. They don't generate useful enough documentation on their own, except maybe if you use all of the tags, like <example> which is at least as much work as doing the above I think.
Edit: In practice, at least in my own experience, I find that class diagrams aren't very useful, UML can be especially for large systems, but the way it's used it often isn't.
To err is human. Fortune favors the monsters.
|
|
|
|
|
honey the codewitch wrote: I find that creating a technical article not only makes it easier for people to get their feet under them using it (at least I think so), it also helps me later to flesh out an outline of the major components of the software, which then evolves into a table contents. Finally that becomes master documentation.
I agree 100%.
I'm always interested in how to communicate technical details to an audience without confusing or boring them. What makes it better?
Mostly, I mean shorter when I say better.
After reading numerous brain books I discovered something that is interesting.
Almost everything you know & think about has been learned through Context.
The easiest example here is you need to remember a list of 15 things.
If you simply attempt to remember them you will get 5 +/- 2.
However, if you build a story around those 15 things it is quite likely you will remember them. Humans need context.
And so it is with documentation (IMO). A lot of documentation is simply a list of technical details.
However, if you provide Context -- think of communicating how the thing works as if it is a story to a listener who is somewhat interested.
This does generally mean documentation is longer but if you do it right it will be so much better it will seem shorter (in most cases).
|
|
|
|
|
If you're talking about documentation similar to Mozilla's and Spring's then I'm 100% on board 
|
|
|
|
|
IMHO it depends very much what you describe in '///' If it is only the list of parameters then it would be useless.
I saw a lot of such documentations for APIs... but the list of parameters I see allready in the code.
|
|
|
|
|
I created "leveled" Data Flow Diagrams (DFD's) for all the financials of Company XXX which were turned over when everything was outsourced. In effect, you're reverse engineering. Code comments won't do it if you expect someone else to get a higher understanding of how things work; which is usually what's needed first / most. With a high level view, it's easier to decide at what point to stop with the details.
"Before entering on an understanding, I have meditated for a long time, and have foreseen what might happen. It is not genius which reveals to me suddenly, secretly, what I have to say or to do in a circumstance unexpected by other people; it is reflection, it is meditation." - Napoleon I
|
|
|
|
|
I'm using Sandcastle Help File Builder myself, which generates documentation from comments in code.
It also allows to add custom content using the "MAML" language, but that has a steep learning curve.
Here are some other options: documentation-tools-for-net-developers[^]
|
|
|
|
|
Documentation of "what" (such as that generated by Doxygen) is happening is not always that useful. I find an explanation of "why" is much more helpful. Unfortunately there is no tool that can do that for you.
- I would love to change the world, but they won’t give me the source code.
|
|
|
|
|
Best documentation I see explains "Why?" Ask why the code exists, and sometimes for why the design was used vs other alternatives. The level of explanation depends on the complexity of the project.
|
|
|
|
|
The why often depends on how something fits into the bigger picture. The rationale can be anything from architectural to it had to be done this way to mesh with some PoS that we didn't have the time to rewrite. These things would usually be described in a the HLD, but the discipline to make those living documents is rare.
|
|
|
|
|
I'm in the situation now where I have to take over an array of projects without any help from the original programmer.
Here's what I want to know: HOW DO I RUN THIS STUFF LOCALLY!?
Any configuration I need to know about?
Also, how do I deploy!?
Do I have to copy some files, run database scripts, do a VS deploy, run a pipeline?
When "the business" talks about an application, what Git repositories do I need?
Also, are there any projects that should be changed together, which is not obvious from the code (for example, because they share a database dependency).
I'm rarely interested in documentation about the code.
I can read the code.
I can click a button/call an endpoint and set a breakpoint to follow through the code.
The things I want to know are the things I can't know without knowing (this sounds more profound than I had meant it to be  ).
|
|
|
|
|
That I have documented for the most part on one of the projects so far
We're putting some effort into documenting a lot of our processes and info now. Previously it had been in emails, some documents, etc. but now it's getting combined into Confluence. Finally... now that I'm leaving
modified 14-Apr-22 1:00am.
|
|
|
|
|
Pretentious: Good code doesn't need documenting, it is self explanatory.
Reality: I am lazy
In practice, I try to minimize in-code documentation to reduce clutter and limit it to "magic" code, where I have used an unconventional technique that might confuse a lesser mortal.
Nothing succeeds like a budgie without teeth.
To err is human, to arr is pirate.
|
|
|
|
|
absolutely agree. Suffering with this right now.
Charlie Gilley
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
Has never been more appropriate.
|
|
|
|
|
I think it's more your "experience" knowledge that will be useful for the next person - so I would argue that a document explaining the domain knowledge and why certain APIs and code work they way they do may be of far more use than code documentation.
I say this because I spend a lot of time reading code and the most useful information I get from other developers is information like "yeah, this is what can go wrong and this is how the customer uses the API".
“That which can be asserted without evidence, can be dismissed without evidence.”
― Christopher Hitchens
|
|
|
|
|
You asked for advice and are getting a whole range from a diverse set of viewpoints!
I will say that some of the most important things to document are:
1) Dependencies/Build/install/setup
2) Validation. How do you test/validate (or who else is responsible). Built in Testing?
3) Deployment Guidelines
4) Risks, Previous issues, and the most problematic code
5) The overall architecture, and a "Metaphor" for how to view the system
6) Any indication of how they might locate something (or the logic of the structure choosen)
Look at it this way. Assume YOU were walking in on day 1. The new developer to replace the previous.
You have NO CLUE what this system does. How to build it or test it. If you are even using the latest source.
Your confidence is in the toilet, and they need you to get up to speed.
What type of information would increase your confidence, and get you able to BUILD and VALIDATE the build the quickest?
Then what information would allow you to NAVIGATE the code logically?
Finally, assuming you only fixed a Typo... On your first day. What would make you feel okay to publish that change???
Document those things. NOT the code, per se.
IMHO
|
|
|
|
|
Sometimes the best coding dramas are written in the comments. (aside from some funny jokes)
But sometimes a good laugh is the best way to make people remember the oddities of certain hacks or unusual algorithms.
If I ever see Windows ME's source code, I'd expect to see shitfaced comments all over the people (evidence of Balmer's Peak)
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
