|
Let me know if it would be better to create a survey....
background: I handle building a multi-part application that crosses multiple development tools, custom processing, etc. Basing my "automation" work off of some work by another developer, I essentially have a dozen MS-DOS scripts that build each part and then assemble the whole. Now, IMHO, MS-DOS scripting just blows, and I'm being generous. Sure, it gets the job done, but calling it archaic is too generous. So, I'm looking to implement something a bit more modern that will work effectively with a nightly build process.
So, what do you use? MS-DOS script users need not reply 
I'm looking at SCons now, but thought I'd get some feedback from CP.
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
I used to work on similar projects and used make and makefiles. They are exceedingly flexible, customisable and work with just about any directory structure. The learning process (as with all) can be a little steep but time spent learning it does pay dividends. I have done something similar on Windows with Visual Studio projects, but on a much smaller scale.
|
|
|
|
|
I perceive you are an old fart (no offense, I am too). I used to use MMS (DEC) - very similar to make. I'm headed in that direction, but I'm looking to put a little more intelligence in it.
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
charlieg wrote: I perceive you are an old fart
More like a very old fart ... I first started programming in machine code on Leo Computers Society. Leo 3 photos[^] (first four pictures in the second row) in 1966.
|
|
|
|
|
charlieg wrote: but I'm looking to put a little more intelligence in it.
Not sure what that means but you can add conditional build semantics to make files if that is what you are referring to.
|
|
|
|
|
We're using Jenkins[^] where I work.
Latest Article - Slack-Chatting with you rPi
Learning to code with python is like learning to swim with those little arm floaties. It gives you undeserved confidence and will eventually drown you. - DangerBunny
Artificial intelligence is the only remedy for natural stupidity. - CDP1802
|
|
|
|
|
A daily build script for your project[^] 
Bastard Programmer from Hell 
If you can't read my code, try converting it here[^]
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.
|
|
|
|
|
Touche'  I'll review it.
Thank you
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
It has been a while since that, and I heard some good things about FogBugz. You may want to browse around a bit, after reading the tip on how to do it yourself.
Bastard Programmer from Hell
If you can't read my code, try converting it here[^]
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.
|
|
|
|
|
Our build process used to be exactly as you describe: a mish-mash of batch files, VBscript's, and a few odds-and-ends executables thrown in for fun. I looked at a few of the existing build tools before writing a Windows service written in C#. Most of the public tools were preferentially biased toward a specific tool-chain or target operating system. The other problem was customization - you had to learn their "paradigm" in order to adjust the process to your requirements, and you could spend a lot of effort getting around paradigm features that weren't needed or appropriate.
My service handles everything: get source from source control, compile, move files as needed to compile installers, create install images, and finally generate .ISO's to archive the build. The service communicates with our in-house network tracing tool to display progress and to let users cancel or pause builds. Builds for specific products are implemented as classes deriving from a base 'build' class. Product build classes can override steps in the build process as needed. New products and branches can usually be added in minutes, rather than the days it used to require. The whole thing is vastly easier to debug and maintain than the original wretched hive of scum and villainy  .
Software Zen: delete this;
|
|
|
|
|
Automated builds
To me that means getting F5 pressed without having to explicitly do it myself.
You mean there's more to it?
|
|
|
|
|
lol, yeah, I wish. Our release process can become something like "We need to ship something to test tonight." This might occur at 9am or 5pm. Manually performing all the steps is a recipe for disaster. The entire point of post is I just in fact did this - screwed the build up.
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
It turns out writing software is easy.
It's the build systems that are downright horrifying.
|
|
|
|
|
PowerShell is the modern way to write scripts, according to Microsoft. That might be an acceptable option. There are also tools for developing PowerShell things in VS2017. I would take that route before resorting to writing custom tools.
"They have a consciousness, they have a life, they have a soul! Damn you! Let the rabbits wear glasses! Save our brothers! Can I get an amen?"
|
|
|
|
|
I've looked at powershell briefly. Maybe it's just me but it reminds me of someone on crack cocaine trying to write a "scripting" language.
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
As per Marc - Jenkins
We're a c++, c#, Ruby shop - Jenkins calls wrapper scripts and/or nmake, MSBUILD, Rake etc etc - its quick to get up and started
One of the things you may need to consider is build/test pipelines and whether parts of the pipeline need to be run on different servers (we typically run build server(s) and test server(s), Jenkins works well for this, the test server(s) are 'pooled' where we need to, dedicated for other builds)
|
|
|
|
|
Based on your comments you are going to need more than just a normal 'build' system.
A build system creates the artifacts. But you are also looking for automated deployment.
And I suspect you might have distinct and perhaps different deployment targets (QA and prod) for example.
That complicates the process.
And it my experience doing that will require a mix of tools.
From your post I suspect you haven't even scoped out the actual use cases. Which should be your first step.
Let me provide some use case scenarios
- Allow a way to run unit tests (not on the developer machine.)
- Allow a way for a developer to run the full deployment on their machine, so they can test the full process.
- Allow a deployment to a QA box.
- Allow a deployment to a QA stack. Different than the above since it might include a cloud target.
- Allow a deployment to create (important) and deploy to a production clone target.
- Allow a full build, versioning to specified target. So a release candidate for QA or production.
- Allow a deployment to the Sales Demo box.
- Allow a deployment to a Professional Services box.
- Allow a hotfix build (start with a versioned build and apply a branch)
Your use cases presumably would be different and you might not want to build all of those now but you need to understand all of them now so you don't make any decisions that complicate or even preclude the scenarios you need.
|
|
|
|
|
baby steps dude, baby steps 
I live in an embedded world, so there is no cloud- it's a lump of hardware sitting on a desk. It may or may not be production hardware.
I just want to know if everything built correctly. For example, I just did a full build, hardware A is at 9.346, hardware b is 9.344.
bad
Right now I build with ms-dos scripts and it's as painful as an in-grown toe nail.... but your point is valid. There just isn't the resources to support all of that....
Charlie Gilley
<italic>Stuck in a dysfunctional matrix from which I must escape...
"Where liberty dwells, there is my country." B. Franklin, 1783
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
|
|
|
|
|
https://www.nytimes.com/2019/02/01/nyregion/standardized-testing-teachers-students.html[^]
How do you test a teacher's performance? Not by seeing if his students have learned anything, obviously.
So our suffering at finding that highly-qualified programmers don't know how to code will be shared among other professions.
Contrary to the popular old saying, I don't think that will halve it.
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
|
The problem is that teachers don't want to be performance tested: they know that a good proportion of them are poor teachers and suspect that they would not be in the "excellent" group if tested by any actual metric.
Though to be honest, it's difficult to test and assess teachers with a broad metric like that: a ghetto school in Bedford-Stuyvesantis going to need a different type of teacher to a private school teacher near the Upper West Side ... "bullet proof" is probably in the job description for both, but for different reasons! 
Sent from my Amstrad PC 1640
Never throw anything away, Griff
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
My wife is an elementary gifted teacher. Best teaching job there is in my opinion. Those are the kids who are actually eager to learn and she asks me to do cool things like build a trebuchet for her Middle Ages unit. The kids come to my wife one day a week, like 2nd on Mondays, 5th on Wednesdays, etc. She works with them 1 day a week and the kids are scattered among 3 or 4 classes per grade level. The gifted kids may make up 5% to 10% of the entire school.
She is on one committee in the school dealing with test scores. The scores for each class are published in the local paper as an average of all the students in a class. Each kids gets a score from 1 to 5 on some test. The other members of the committee were pressuring my wife to work more with her kids to make sure they all got fives because that would raise the score, instead of focusing on the majority of the kids who were making 1's and 2's to get them up to 3's and 4's. 
|
|
|
|
|
MarkTJohnson wrote: The other members of the committee were pressuring my wife to work more with her kids to make sure they all got fives because that would raise the score, instead of focusing on the majority of the kids who were making 1's and 2's to get them up to 3's and 4's.
But, for her personal appraisal, it doesn't matter if she took 3s and 4s and made them 5s, or 1s and 2s and made them 3s.
Either way, if she managed to do it with a lot of them, she did a good job, so her personal appraisal, based on how much children learned, should be a good one.
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
|
My point was that they were focused on the small percentage of gifted kids who my wife sees one day out of 5 rather than the other 90% of the school that those teachers have all 5 days of the week.
Being the gifted teacher my wife's name does not get reported in the paper since she isn't one of the grade level teachers. Her kids were going to be 4's and 5's anyway.
|
|
|
|
|
MarkTJohnson wrote: My point was that they were focused on the small percentage of gifted kids
That's what happens when the results are used as a political football, which is about as ethical as everything else in politics.
Good teachers should be rewarded and trusted, and the only real way to judge if a teacher has done a good job is by looking at the children they teach -- but that should be local, not national, the same way everyone has their personal performance assessed.
Children's results are personal, and can be used as indicators of how well they've been taught; using them as statistics to push political agendas stinks.
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
|
OriginalGriff wrote: a ghetto school in Bedford-Stuyvesantis going to need a different type of teacher to a private school teacher near the Upper West Side
Sure, but a personal appraisal is about "Did this one person do a good job?", so that will scale according to the expectation of the school as a whole.
A good teacher will teach more, regardless how high or low the baseline, and how much is considered "more" and how much "less".
You can't blame the teachers if a school in a bad area gets poor national results, but you can tell which of the individual teachers is better by how much they manage to teach to the students.
"How much our students learn doesn't matter" is a crock of sh1t, because it's the only thing that matters.
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
