|
Wordle 337 4/6
⬜⬜⬜⬜🟨
🟨🟩⬜⬜⬜
⬜⬜⬜⬜🟩
🟩🟩🟩🟩🟩
Software rusts. Simon Stephenson, ca 1994. So does this signature. me, 2012
|
|
|
|
|
Wordle 337 5/6
⬜⬜⬜🟨⬜
⬜⬜⬜⬜🟨
🟨⬜🟨⬜🟩
⬜🟨⬜🟨🟩
🟩🟩🟩🟩🟩
|
|
|
|
|
Watching system usage, it blows my mind how much differently programs work. Right now Vivaldi has over sixty tabs open (way too much, even for me), and Task Manager shows ~1 GB. Firefox only has 3 tabs open, and it is at 1.8 GB. Even though 2 of those are for Facebook, the difference still seems staggering.
Just for curiosity's sake, I shut down FF, and reopened it. Without clicking on any other tabs, it still shot up to the same memory usage as Vivaldi. Clicking on the other tabs took it up to 1.2 GB, and scrolling FB quickly added another 100+ MB to it.
I see many people praising FF, but I still remember the old problems with it, including ever-expanding memory usage. It doesn't seem like the underlying memory usage has really been addressed, even after all these years.
Out of curiosity, I opened Vivaldi in Windows Sandbox, with the same 3 tabs. The memory usage was half that of FF. Firefox, I do wish you well. But it would be nice to see better engineering behind the scenes.
|
|
|
|
|
Meanwhile, Chrome is like "Oh, I see you have 32GB of RAM... It would be a shame if something happened to it..."
|
|
|
|
|
In the old days, Chrome was an absolute pig. I now have a Chrome window open with only 7 tabs, and it is far more memory efficient than FF would be, at only 472 MB. Chrome has done wonders with memory efficiency compared to FF, in my experience. But it seems the folks at Vivaldi have taken Chrome's base and polished the heck out of it. They have performed an absolutely impressive feat, especially with the additional tools they have given us, like command chains.
|
|
|
|
|
What irritates me about FF is that they never fix security vulnerabilities. We use Rapid7's InsightVM at work and even after updating Firefox, the same security vulnerabilities reappear. Chrome and Edge both fix security vulnerabilities. From this perspective I have banned FF from our work machines except for our marketing folks who have to test web-site updates in as many browsers as possible.
|
|
|
|
|
Sad to hear that. I like the idea of Firefox being an independent web browser, but things like this make it hard to take seriously.
|
|
|
|
|
My current job deals with a multi-gigabyte code base, a very complex application in the Finance sector. It's brilliantly designed, in my opinion, and uses OO to great effect.
However, there is zero internal documentation on how it works, how the data flows through the different layers and what components do what.
I have had to ask (over Teams) for every bit of knowledge I have about this application, and there are still places where I'm completely in the dark.
If I had to do it again, I'm not sure I would have taken this job, due to the frustration level. How important do you think internal documentation is, and would you take a job with a company that has none?
The difficult we do right away...
...the impossible takes slightly longer.
|
|
|
|
|
“Programs must be written for people to read, and only incidentally for machines to execute.” - Hal Abelson
Mircea
|
|
|
|
|
Not sure what you mean. 
The difficult we do right away...
...the impossible takes slightly longer.
|
|
|
|
|
Programs without appropriate documentation might as well not exist. They cannot be maintained and, sooner or later, they have to be replaced. Programs with documentation are means to transfer knowledge between people. And yes, they also happen to be executed by computers 
In other words, documentation is critical in my opinion.
Mircea
|
|
|
|
|
I'm working on a much smaller scale project, but it also has this lack of documentation / explanation. It makes it difficult for new team members to get started on it. Documentation would have helped.
Now that you have the knowledge you can create some documentation 
|
|
|
|
|
Sadly there was no or completely infsufficient documentation in the companies I worked for.
Although it is easy to blame the developers, the real culprits I think are management that keeps asking for more bells and whistles instead of allowing time to document things properly.
|
|
|
|
|
Such undocumented code is, unfortunately, more the rule than the exception.
The only thing I can suggest is that as code is modified (maintained / upgraded), you write comments in the code describing what the code does and why. Even a comment header may be useful for those who must follow after you.
I realize that some code bases require that changes be minimal and address only the required change. In that case, you should provide an external document (ensuring that everyone knows where it is), and write your observations in there. This is less ideal that in-place documentation, but will ensure that whatever information you recovered from your colleagues will be there for "future generations".
Freedom is the freedom to say that two plus two make four. If that is granted, all else follows.
-- 6079 Smith W.
|
|
|
|
|
The heck... My initial reply here was for the thread above this one 
Anyway, I am one of those developers who does not write documentation (although I should).
However, what I'd like to see in documentation is:
- Where can I get the code?
- How can I run it locally for debugging purposes?
- How do I commit it to source control (dev branch, merge request, directly on main branch, etc.)?
- How does this application fit in the landscape (is it stand-alone, does it depend on other services, do other services depend on this one)?
- How can I deploy it to various environments (and what environments do we have)?
- Does it depend on any in-house packages and if yes, where can I find those packages and how do I update them?
- Do I need to know about any secrets, passwords, ID's, servers, etc.?
What I'd absolutely NOT want to see in documentation: how does the code work?
Code changes constantly, so I'd have to constantly change the documentation and I'd have to know where I should change the documentation, which is even harder than changing the code.
In code I can check where I use a certain variable, I can't do that in documentation.
So in order to update technical documentation I'd have to very carefully read it every time I change something (and I still wouldn't know if I missed something).
Any documentation I've ever read about code was outdated and did not reflect what the code actually did.
I can follow code using the debugger, so that's its own documentation and never lies to me.
A well-placed comment can work miracles there (like, I have one property that I cannot rename, I have two seemingly separate lines that should be executed in a certain order, and I have a use case for Thread.Sleep).
That said, I now have a client that has a small ASP WebForms layer that only puts records in a database.
The database is then read by some no-code background process.
For starters, I can't know which background process uses which tables without checking ALL background processes (which would take you some hours).
Every process is very hard to follow, has a lot of steps, and A LOT of tables, views and stored procedures.
It's mostly "read the data from some table (or view), send the data to a web API, store the results in yet another table, run a stored procedure that processes those results and puts them in a third table..."
I'm not sure documentation could fix it though 
Documentation should never be a way to fix broken code (or no-code, in this case) or a broken architecture.
Documentation can be really useful for manual processes, like enabling a client or user to use OAuth in your application (if that's a use-case you have, of course).
More like a sort of internal FAQ, what should I do when a customer asks me to...?
That's more functional than technical documentation though.
Documentation has never been a reason for me to join or not join a company.
If a team of six developers can help me with any of the above questions that's almost as good as documentation (until all six decide to quit ).
|
|
|
|
|
Can I respectfully disagree?
First let's clarify what I call "documentation". There is no generic "documentation" but documentation for a certain purpose. To make an analogy with the auto industry, you have the user manual that you get with any new car, than you have a service manual that mechanics use and you have design documentation that stays in the factory. They are not interchangeable and none of them replaces the other ones. You don't want to look at the service manual when you buy a new car and the mechanic has no use for the users' manual (you'd hope he's seen that type of car before). And of course you don't get to see the design documentation that manufacturers guard jealously.
Pretty much the same applies to software. I think the OP was talking about design documentation, or the lack of thereof. It should explain the whys and tell little about the hows. Presumably the purpose of the system and the design don't change that often as to make it constantly obsolete.
What you seem to refer to, is what I call "user manual": how to get the code, where are the passwords and so on. These are the nitty-gritty details that are very helpful but don't explain why the system was designed in a certain way nor how it was implemented.
Finally, for the "service manual", the one that explains how the design comes to life, comments are just perfect. They are right there on the shop floor when the mechanic needs to fix something and can be updated right away when something changes.
Mircea
|
|
|
|
|
It doesn't really sound like you're disagreeing, just explaining things differently, or thinking about other documentation.
We're developers and I assume the documentation Richard is referring to is documentation for us, developers.
I'm also assuming he's not talking about comments in code, because he'd call them comments and not documentation.
Mircea Neacsu wrote: I think the OP was talking about design documentation, or the lack of thereof. It should explain the whys and tell little about the hows. Presumably the purpose of the system and the design don't change that often as to make it constantly obsolete.
That would be my following two points:
- How does this application fit in the landscape (is it stand-alone, does it depend on other services, do other services depend on this one)?
- Does it depend on any in-house packages and if yes, where can I find those packages and how do I update them?
Those two points would also answer Richard's question of "[...] how it works, how the data flows through the different layers and what components do what."
You're a bit more detailed, I think, wanting to know how an application is build in terms of interfaces, base classes, etc. which could be a useful addition if this isn't directly clear from the code.
For example: "For process X we're using IImporters and every client has their own implementation of IImporter which can be found in ... and depends on ... and is chosen/instantiated as follows ..."
Overall, I'm not very concerned with the why of the system.
It's here and people use it for one reason or another, that's a given.
Knowing globally what it does can help in finding and debugging functionality, but it's not a requirement.
That changes when you're tasked with thinking about how to make the program better and more useful for your customers, but that's not usually the task of a programmer and I wouldn't want to find it in the documentation that's supposed to help me run, debug and change a system.
Mircea Neacsu wrote: What you seem to refer to, is what I call "user manual": how to get the code, where are the passwords and so on. These are the nitty-gritty details that are very helpful but don't explain why the system was designed in a certain way nor how it was implemented.
Depends on your point of view.
If a user asks for a user manual, they don't want the code and they're not even allowed to secrets.
A software user manual, to me, would be "to create a new account, find "accounts" in the menu and click on the plus icon ..."
That could be very useful to a developer when a user reports a bug in account creation, but usually you'd ask for steps to reproduce anyway
|
|
|
|
|
I have been considering two alternatives for activities when I retire: Either, I could clean up messy, amateurish Wikipedia articles (maybe even add information, if the article is within my field of expertise. I don't see any possibility of running out of Wikipedia articles that need a cleanup.
Or, I could start working through open source projects, not contributing a single line of code, but adding documentation, from the very top level down to the individual methods, data exchange paths, data structures etc. Or, the intention would be to make neither code extensions or corrections, but I know very well from my documenting my own code (and partially that of others), that it serves perfectly well as rubberducking. When you have to give others a clear and concise description, you discover lots of things that passes by as long as you rush on to implement that new exciting functionality.
I guess that I will stick to Wikipedia. Revealing a messy architecture, inconsistent data structure designs, lack of thread protection where there should be, etc. in a clear and concise manner is bound to lead to a lot of people being upset for me picking on their code. I think that lots of programmers (both of closed and open source code!) fear a view from the outside on their program design. In code reviews, they are used to others reading their conditional statements and loops; that part is OK. But when a system has grown "organically" for five years into a terrible mess of pieces from the method level and upwards, it is rarely subject to a code review. And that is where the problems usually are found. And fixing them requires far more effort than fixes on the code level. No one wants to be told that their architecture and module design have deteriorated into a can of worms.
Well, it rarely is that bad, but most code bases that I have been in touch with have had so many deviations from what was once the original architecture that it negatively affects both reading and maintaining the code. Practically always, I hear "Yes, we know, but ... We hope to get around to clean it up ..." They would not want to have that in writing. They want to pretend that they have gotten that round tuit, even though it isn't the case.
I guess fixing up Wikipedia articles will make fewer people upset.
|
|
|
|
|
It's a curse of knowledge thing.
The user domain and coder knowledge domains don't overlap, so the user thinks they have full documentation WITHIN THEIR DOMAIN (and they probably do), but that's a world away from from it being transferable knowledge and understanding.
Usually, any documentation is some hasty ramshackle bridges from the swampy margins of one domain toward the other, with gaps at the most awkward bits. It's not documented (paved) because almost no one passes that way. It's like finding the old Route 66 (so I'm told).
Have you documented your code (rhetorical..) to a level so that the finance folk could debug it?
|
|
|
|
|
My first job where I did programming was when I was an engineering liaison for NASA Langley. We never embarked on a large project without first doing all of the documentation. All. It made perfect sense to me, because no person working on writing the code was in the dark about what he was writing and where it fit into the whole, and it left a trail behind for maintenance to use.
Having taken 25 years off to raise children (mostly - I did continue to do web development), I found that this documentation-first idea had been abandoned by the majority of the programming world. I can only assume that it was abandoned for the sake of development speed, and that some creators think that their creativity will suffer from the delay and tedium of creating the documentation. I think it's a bad idea. I mean, I can read code and grep as well as the next guy, but I could have been productive much faster and an expert on the codebase in no time if there'd been documentation, not to mention that the code itself would have likely been more consistent and clean.
jmo
|
|
|
|
|
You haven't heard the "code should be self documenting" crap? I had a manager who wanted comments stripped out precisely because of this.
I will sometimes leave code that has been commented out in just to remind future me, "Oh yeah, we tried that before and it didn't work."
I’ve given up trying to be calm. However, I am open to feeling slightly less agitated.
|
|
|
|
|
For anyone with good knowledge of the language, code is self-documenting, but the document is wildly incomplete, entirely lacking an indexing system (apart from grep which is woefully inadequate), and spread across disparate files in a filing system that is likewise lacking indexing.
I know that part of the appeal of frameworks is to provide some kind of indexing for finding what you need to work on, but frameworks also come with both bloat and inescapable opinions about structure, and not all projects you will come to will be built on one (the one I work on is not, for which I am grateful).
|
|
|
|
|
I once took over a codebase that was in the form of hundreds of UNIX shell scripts in which there was not a single line of comments in any of them. This was exactly at the moment that things were rapidly going to hell in production, and the problem was somewhere in this group of scripts. I reached out to the 'programmer' of the scripts who said that the code itself was the documentation and that (he knew that I was a very good shell programmer) I should be able to figure it out. While it is true that I could unwind the code eventually, we were in a crisis situation where I needed to quickly identify the specific line of code that was causing problems. There was no time for crawling through code, and that is the whole point.
If there is a high urgency, there is no time to 'learn the code'. In that case, a simple, one line sentence at the top of each script--'this script does x'--would have helped enormously. SOME documentation is ALWAYS needed.
Cheers,
Russ
|
|
|
|
|
I agree a lot with what @Sander Rossel[^] and trønderen - Professional Profile[^] had to say. I would add to that that a good high-level architectural diagram can do wonders for quickly learning how the code works. In Psychology, it's called a cognitive map. If a person has a map of the big picture, then it's very easy to fill in the details and know how it all goes together. If you only have the details (the codebase), then you have to create the big picture. If creating knowledge was easier than absorbing it, then school wouldn't exist.
Bond
Keep all things as simple as possible, but no simpler. -said someone, somewhere
|
|
|
|
|
I learned the lesson of good documentation early in my career when I wrote a quick and dirty assembly tool thinking I would only use this once and throw it away. Long story short, 6 weeks later I need to change something and ended up having to spend time reverse engineering my own code. Since that time I have always documented my programming.
Sadly, I can only recount 2 other projects I have done in 35 years that documented anything.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
