Originally posted June 02, 2018, 02:07:40 PM; partially reposted to assist splitting the topic. This section of the original post is about specific issues in THLeaderH's changes, many of which had been raised before. The remaing paragraphs can be found
here.I was looking through these changes last night, and I noticed that many of the issues mentioned in the Standard thread for this patch are still present in this thread. I've made some changes to your code, which James has now incorporated into master, so you should be able to look at them fairly easily. In no particular order:
(Also the patch files of github seems uneccessary large since a lot of identical lines (maybe with whitespaces) are contained within.)
This was my first observation too. In fact, while the full diff of your changes (using standard git diff settings) is over 10000 lines long, if I tell git to ignore whitespace changes then the diff is less than 5000 lines long. While it might be good to fix whitespace inconsistencies, this should always be done separately to other changes. The reasons for this are that it allows people to inspect and understand your changes, and that it makes it easier for people to understand the history of the code.
To take one particular example, if I look at boden/wege/weg.cc, there are a lot of changes to whitespace mixed in with actual code changes, all under the commit message "bring from standard". Ideally the code changes would have been separate to the whitespace changes, and that commit message (at least read in isolation) is quite misleading, as these changes were not taken from standard, but from a patch on top of standard.
Logic incomplete and even flagged with TODO's, yet still left.
I spotted and removed one out-of-date TODO relating to version numbers (which is fairly reasonably something to leave in until James merged it). However, there are still some others, among which this one in bauer/wegbauer.cc particularly worries me:
// TODO: not edited since it's quite different from that of standard...To me, this TODO suggests that you haven't bothered to port part of the code because it was difficult. I hope this is not the case, but some reassurance here would be appreciated.
Maybe try using this new fangled keyword called 'else'. if(a), if(!a) yeesh.
It took me less than a minute to find an instance of if(a) {stuff} if(!a) {stuff} in your changes, and even though it might take you longer if you're not as familiar with the relevant git commands, you should still be capable of looking at your changes and reading through for things like this. I think I replaced all the instances of this particular non-pattern, but since I haven't read the patch through line-by-line I cannot be certain.
refleshing. what? I can't even...
I think Turfit could have written this better, but I'm pretty sure his gist is that "reflesh" is almost certainly not the word you meant. My guess would be that you meant "refresh" instead, but even that seems like a poor name choice, with "refresh_vehicle_images" probably being a better name for the method in overtaker_t. (I would also make this method take no arguments, and move the check into overtaker.cc).
I also removed the "reflesh" method from vehicles because (as it was implemented) it was duplicating a lot of code that already existed - see my changes for (in my opinion, at least) a better way of doing that.
Another thing I've noticed that could be improved in your code is a lot of places where your manipulation of booleans is unecessarily complicated, or you store intermediate results that don't need to be stored. In one example that I simplified last night, you use about 5 lines of code with an if branch in the middle where you could have used a short boolean expression. A couple of lines later you create a whole new overtaker_t variable just to test if it is null. Storing the object suggested that you would use it again later and that the null check was partly to avoid null dereference errors; in fact I realised after a bit that the null check was the entire usage of the object.
Out of interest, how did you solve this issue? I came across it this morning while reading through the Standard thread, and didn't notice any mention of how you resolved it.
https://forum.simutrans.com/index.php/topic,16659.msg165492.html#msg165492Edit: Four paragraphs extracted to a new topic - see
here.
I also know that there are many more changes that could (and should) be made to improve the code. I'm not going to list every single change that should be made, because that would take me days or more likely weeks, and then I might as well make them myself. (See also 'code smells'.) That is the point Turfit was making - we can tell you the sorts of issues we see, and we can maybe give you some examples, but you then have to be able to extrapolate from this and seek out other instances of these issues yourself. For example, you complained "Please make your issue posts as concrete as possible. For example of if ~ else, please point out at least 1 line that is malformed." Did you even try looking yourself? As I said, it took me less than a minute of skimming through your changes to find an example of this; it shouldn't have taken you too long to find one yourself, even if you were unlucky and looked in the wrong places first. If you can't look for and find issues yourself, even after you are told what those issues look like, then the only alternative is that someone else rewrites your patch for you. That shouldn't be necessary.
With your knowledge and understanding of the code, you should be able read through it and rewrite bits of it more easily than me, because everywhere that something is written out poorly (e.g. with duplication, unnecessary complexity or vague of misleading code/comments) you already know what it's supposed to do. Fundamentally, what I (and Turfit, and prissi, and Ters) want is for you to read through the entire patch, line by line, and section by section, a couple of times, and think about how you could make it easier for other people to read. I'm sure James will be glad to incorporate such changes, and that is also the only way you're going to get this into Standard.
(James: I think this should enable you to split the topics - posts 34 and 37-41 being the new topic)
