> Except that in software 1000's of groups have already dug 1000's of canals and yet the next group that will build a canal will insist on doing it their way.
This is why a prudent developer/team would attempt to leverage frameworks/plugins/existing API's and libraries vs reinventing the wheel. Leveraging well-vetted and popular resources should cut development time considerably.
> This is why a prudent developer/team would attempt to leverage frameworks/plugins/existing API's and libraries vs reinventing the wheel.
The problem here is that those frameworks/plugins/existing API's are all moving targets, their life-span is very likely substantially shorter than the life-span of your project.
Well vetted and popular unfortunately does not equal 'will have staying power'.
That hasnt been my experience. Most of the frameworks I use last much longer than my tenure at a particular position. Are you saying you would really write your own django? jquery? react? chef? git? I would argue the reason some startups can now move blazingly fast with relatively few developers is by leveraging open-source in a smart way.
> Are you saying you would really write your own django? jquery? react? chef? git?
No, I think it should be fairly obvious that that would not be my choice, rather the opposite.
But why django? Why not rails? Or symphony? Or Yii? Or Spring? and so on, why git? and not subversion, or tfs or mercurial? and so on. For every one of those choices there are 10's of possible choices and none of those will be something akin to an industry standard. And so we muddle on. Well, at least git seems to become slowly the standard in revision control, but because git was yet another fresh start project a lot of the lessons learned from other revision control systems that were not readily apparent to the git author(s) had to be re-learned the hard way, resulting in a lot of delay and frustration (and to this date the git command set strikes me as ill thought out).
Leveraging open source in a smart way is exactly the right thing to do, now if only there would be some kind of mental penalty to start a new project when an alternative is already available which could use some TLC.
This is true for the lifespan of that framework. No software anywhere has the devoted love of its adherents that Borland's Delphi holds. So where it it now?
And I have literally replaced piles of "well-vetted and popular resources" with better-vetted and less popular resources in matters of weeks - after all, the knife-fighting over features was done, and all that was wrong was that the framework had some massive hole.
This is why a prudent developer/team would attempt to leverage frameworks/plugins/existing API's and libraries vs reinventing the wheel. Leveraging well-vetted and popular resources should cut development time considerably.