index.md (6336B)
1 Stuff that sucks 2 ================ 3 See the [philosophy](//suckless.org/philosophy) page about what applies 4 to this page. 5 6 Bigger topics that suck: [systemd](//suckless.org/sucks/systemd), 7 [the web](//suckless.org/sucks/web) 8 9 Libraries 10 --------- 11 These libraries are broken/considered harmful and should not be used 12 if it's possible to avoid them. If you use them, consider looking for 13 alternatives. 14 15 * [glib](http://library.gnome.org/devel/glib/) - implements C++ STL on top of C 16 (because C++ sucks so much, let's reinvent it!), adding lots of useless data 17 types for ["portability" and "readability" 18 reasons](http://library.gnome.org/devel/glib/unstable/glib-Basic-Types.html). 19 even worse, it is not possible to write robust applications using glib, since 20 it [aborts in out-of-memory situations](https://bugzilla.gnome.org/show_bug.cgi?id=674446). 21 glib usage is required to write gtk+ and gnome applications, but is also used when common 22 functionality is needed (e.g. hashlists, base64 decoder, etc). it is not suited 23 at all for static linking due to its huge size and the authors explicitly state 24 that ["static linking is not supported"](https://bugzilla.gnome.org/show_bug.cgi?id=768215#c16). 25 26 Alternatives: [libmowgli](https://github.com/atheme/libmowgli-2), 27 [libulz](https://github.com/rofl0r/libulz), 28 BSD [queue.h](https://man.openbsd.org/queue)/[tree.h](https://man.openbsd.org/tree) macros. 29 30 * [GMP](http://gmplib.org/) - GNU's bignum/arbitrary precision 31 library. Quite bloated, slow and [calls abort() on failed 32 malloc](https://gmplib.org/repo/gmp/file/tip/memory.c#l105) 33 34 Alternatives: [libtommath](http://www.libtom.net/LibTomMath/), 35 [TomsFastMath](http://www.libtom.net/TomsFastMath/), 36 [imath](https://github.com/creachadair/imath), 37 [libzahl](//libs.suckless.org/libzahl) (WIP), 38 [hebimath](https://github.com/suiginsoft/hebimath) (WIP) 39 40 Build Systems 41 ------------- 42 * [cmake](http://www.cmake.org/) (written in C++) - so huge and bloated, 43 compilation takes longer than compiling GCC (!). It's not even possible 44 to create freestanding Makefiles, since the generated Makefiles call 45 back into the cmake binary itself. Usage of cmake requires learning a 46 new custom scripting language with very limited expressiveness. Its 47 major selling point is the existence of a clicky-click GUI for windows 48 users. 49 * [waf](https://code.google.com/p/waf/) and 50 [scons](http://www.scons.org/) (both written in Python) - waf code is 51 dropped into the compilee's build tree, so it does not benefit from 52 updated versions and bugfixes. 53 54 As these build systems are often used to compile C programs, one has to 55 set up a C++ compiler or Python interpreter respectively just in order 56 to be able to build some C code. 57 58 Alternatives: 59 [make](http://pubs.opengroup.org/onlinepubs/9699919799/utilities/make.html), 60 [mk](http://doc.cat-v.org/plan_9/4th_edition/papers/mk) 61 62 Version Control Systems 63 ----------------------- 64 * [subversion](https://subversion.apache.org/) - Teaches developers to 65 think of version control in a harmful and terrible way, centralized, 66 ugly code, conceptionally broken in a lot of terms. "Centralized" is 67 said to be one of the main benefits for "enterprise" applications, 68 however, there is no benefit at all compared to decentralized version 69 control systems like git. There is no copy-on-write, branching 70 essentially will create a 1:1 copy of the full tree you have under 71 version control, making feature-branches and temporary changes to your 72 code a painful mess. It is slow, encourages people to come up with weird 73 workarounds just to get their work done, and the only thing enterprisey 74 about it is that it just sucks. 75 76 Programs 77 -------- 78 There are many broken X programs. Go bug the developers of these 79 broken programs to fix them. Here are some of the main causes of this 80 brokenness: 81 82 * The program **assumes a specific window management model**, 83 e.g. assumes you are using a WIMP-window manager like those 84 found in KDE or Gnome. This assumption breaks the [ICCCM 85 conventions](http://tronche.com/gui/x/icccm/). 86 * The application uses a **fixed size** - this limitation does not fit 87 into the world of tiling window managers very well, and can also be seen 88 as breaking the ICCCM conventions, because a fixed sized window assumes 89 a specific window management model as well (though the ICCCM does not 90 forbid fixed-size windows). In any case, the ICCCM requests that clients 91 accept any size the window manager proposes to them. 92 * The program is based on strange **non-standard window manager 93 hints** that only work properly with a window manager supporting these 94 extensions - this simply breaks the ICCCM as well. E.g. trash icon 95 programs. 96 * The program does not conform to ICCCM due to some **missing or 97 improperly set hints**. 98 99 If you still need some program which expects a floating WM, use it in 100 floating mode. 101 102 Documentation 103 ------------- 104 Somewhen GNU tried to make the world a bit more miserable by inventing 105 [texinfo](https://www.gnu.org/software/texinfo/). The result is that 106 in 2019 man pages are still used and the documentation of GNU tools 107 requires you to run `info $application`. The info browser is awkward and 108 unintuitive and the reason why no one gets further than finding 'q' to 109 quit it. 110 111 Look at GNU tools how to not handle documentation. 112 113 Talking about the suck in enforced HTML documentation, which forces 114 you to open up a 1 Gb of RAM wasting web browser, just to see some 115 eye-candy, which could have been described in the source with some easy 116 way to jump to that line in the source code, is not worth the time. 117 118 The suckless way is to have a short usage and a descriptive manpage. The 119 complete details are in the source. 120 121 Alternatives: roff, [mdoc](https://mandoc.bsd.lv/). 122 123 C Compilers 124 ----------- 125 * [GCC](http://gcc.gnu.org/): as of 2016 it is now written in C++ and so 126 complete suck. Why can't a compiler just be a simple binary doing its work 127 instead of adding path dependencies deep into the system? 128 * [Clang](http://clang.llvm.org/) is written in C++. If you don't 129 believe that it sucks, try to build clang by hand. 130 131 Alternatives: see the Compilers section of the [/rocks/](../rocks) page. 132 133 See also 134 -------- 135 The [list of harmful software](http://harmful.cat-v.org/software/) at 136 [cat-v.org](http://cat-v.org).