top | item 24912780

(no title)

kobigurk | 5 years ago

As someone who is not at all involved in that community, I'd love to hear what are the pits many projects fall into and what does Blender avoid.

discuss

order

Elv13|5 years ago

Out of the top of my mind:

* Gets good enough, but not great. Then most contributor lose interest and the pace slows. It makes the original roadmap impractical and more devs lose interest, which spirals into a dormant project (Inkscape)

* Maintained flawlessly, but no new groundbreaking features get added and there is a slow exodus of users toward new-shiny until it is a perfectly bug-free irrelevant piece of code (most JS stuff, GnuPG, some crypto libs, OpenBSD).

* Maintained into the ground by aggressive/oppressive/elitist/cultist/dogmatic core team followed by a split of the community (pre re-merge of glibc/eglibc, GCC/EGCS, ffmpeg/libav, Gnome/MATE/Cinnamon/Unity, Vim/NeoVim, all of Suckless, most of GNU)

* Fancy/overkill roadmap made at a time when there is plenty of contributors, but so hard to achieve most of them give up before the finish line (KDE4, GIMP 2.10)

* Unresponsive/exclusive community toward new contributors and/or SJW attempts accidentally FUDding a previously functional community (lets not point fingers and start a flame war)

* Unwillingness to break things or pay up the tech debt in order to stay relevant (EMACS, pre-LLVM-era GCC)

* Break everything consistently and often and/or remove your contributions because of a dogma (Gnome)

* The Free-software people and the commercial contributors revolt against each other (OpenOffice/LibreOffice/StarOffice/CollaborraOnline, OwnCloud/NextCloud)

* Misguided attempts at making money by breaking things for your users (GhostScript, eLive, Mepis, Mandrake, every project who moved to AGPL)

* No attempt at ensuring the project can exist without its founder (CentOS, all single-person-army projects out there, all BDFL who refuse to let go when they lose interest)

* Sabotage (ion3, ffmpeg, Cyanogen)

* Suicide by committee (many Apache/Eclipse foundation projects like Apache OpenOffice)

* Make a major incompatible upgrade with no clear migration paths (Python, GTK, KDE, Gnome, AwesomeWM)

* Promise a major upgrade, tell your users to wait for it, then backtrack (Perl)

* Maintain multiple branches of your software forever until you fragment your user base so much you compete with yourself and lose to new-fancy (OpenSSL, pre-rolling Firefox, MySQL, most "enterprise grade stuff")

* crazy rebranding ideas leaving most of your users unaware of the new project (Apache OpenOffice, my own SFLPhone->Inutchuk->Ring.cx->GNU Ring->GNU Jami->Banji fiasco, KOffice).