1====================== 2Design and History FAQ 3====================== 4 5.. only:: html 6 7 .. contents:: 8 9 10Why does Python use indentation for grouping of statements? 11----------------------------------------------------------- 12 13Guido van Rossum believes that using indentation for grouping is extremely 14elegant and contributes a lot to the clarity of the average Python program. 15Most people learn to love this feature after a while. 16 17Since there are no begin/end brackets there cannot be a disagreement between 18grouping perceived by the parser and the human reader. Occasionally C 19programmers will encounter a fragment of code like this:: 20 21 if (x <= y) 22 x++; 23 y--; 24 z++; 25 26Only the ``x++`` statement is executed if the condition is true, but the 27indentation leads many to believe otherwise. Even experienced C programmers will 28sometimes stare at it a long time wondering as to why ``y`` is being decremented even 29for ``x > y``. 30 31Because there are no begin/end brackets, Python is much less prone to 32coding-style conflicts. In C there are many different ways to place the braces. 33After becoming used to reading and writing code using a particular style, 34it is normal to feel somewhat uneasy when reading (or being required to write) 35in a different one. 36 37 38Many coding styles place begin/end brackets on a line by themselves. This makes 39programs considerably longer and wastes valuable screen space, making it harder 40to get a good overview of a program. Ideally, a function should fit on one 41screen (say, 20--30 lines). 20 lines of Python can do a lot more work than 20 42lines of C. This is not solely due to the lack of begin/end brackets -- the 43lack of declarations and the high-level data types are also responsible -- but 44the indentation-based syntax certainly helps. 45 46 47Why am I getting strange results with simple arithmetic operations? 48------------------------------------------------------------------- 49 50See the next question. 51 52 53Why are floating-point calculations so inaccurate? 54-------------------------------------------------- 55 56Users are often surprised by results like this:: 57 58 >>> 1.2 - 1.0 59 0.19999999999999996 60 61and think it is a bug in Python. It's not. This has little to do with Python, 62and much more to do with how the underlying platform handles floating-point 63numbers. 64 65The :class:`float` type in CPython uses a C ``double`` for storage. A 66:class:`float` object's value is stored in binary floating-point with a fixed 67precision (typically 53 bits) and Python uses C operations, which in turn rely 68on the hardware implementation in the processor, to perform floating-point 69operations. This means that as far as floating-point operations are concerned, 70Python behaves like many popular languages including C and Java. 71 72Many numbers that can be written easily in decimal notation cannot be expressed 73exactly in binary floating-point. For example, after:: 74 75 >>> x = 1.2 76 77the value stored for ``x`` is a (very good) approximation to the decimal value 78``1.2``, but is not exactly equal to it. On a typical machine, the actual 79stored value is:: 80 81 1.0011001100110011001100110011001100110011001100110011 (binary) 82 83which is exactly:: 84 85 1.1999999999999999555910790149937383830547332763671875 (decimal) 86 87The typical precision of 53 bits provides Python floats with 15--16 88decimal digits of accuracy. 89 90For a fuller explanation, please see the :ref:`floating point arithmetic 91<tut-fp-issues>` chapter in the Python tutorial. 92 93 94Why are Python strings immutable? 95--------------------------------- 96 97There are several advantages. 98 99One is performance: knowing that a string is immutable means we can allocate 100space for it at creation time, and the storage requirements are fixed and 101unchanging. This is also one of the reasons for the distinction between tuples 102and lists. 103 104Another advantage is that strings in Python are considered as "elemental" as 105numbers. No amount of activity will change the value 8 to anything else, and in 106Python, no amount of activity will change the string "eight" to anything else. 107 108 109.. _why-self: 110 111Why must 'self' be used explicitly in method definitions and calls? 112------------------------------------------------------------------- 113 114The idea was borrowed from Modula-3. It turns out to be very useful, for a 115variety of reasons. 116 117First, it's more obvious that you are using a method or instance attribute 118instead of a local variable. Reading ``self.x`` or ``self.meth()`` makes it 119absolutely clear that an instance variable or method is used even if you don't 120know the class definition by heart. In C++, you can sort of tell by the lack of 121a local variable declaration (assuming globals are rare or easily recognizable) 122-- but in Python, there are no local variable declarations, so you'd have to 123look up the class definition to be sure. Some C++ and Java coding standards 124call for instance attributes to have an ``m_`` prefix, so this explicitness is 125still useful in those languages, too. 126 127Second, it means that no special syntax is necessary if you want to explicitly 128reference or call the method from a particular class. In C++, if you want to 129use a method from a base class which is overridden in a derived class, you have 130to use the ``::`` operator -- in Python you can write 131``baseclass.methodname(self, <argument list>)``. This is particularly useful 132for :meth:`__init__` methods, and in general in cases where a derived class 133method wants to extend the base class method of the same name and thus has to 134call the base class method somehow. 135 136Finally, for instance variables it solves a syntactic problem with assignment: 137since local variables in Python are (by definition!) those variables to which a 138value is assigned in a function body (and that aren't explicitly declared 139global), there has to be some way to tell the interpreter that an assignment was 140meant to assign to an instance variable instead of to a local variable, and it 141should preferably be syntactic (for efficiency reasons). C++ does this through 142declarations, but Python doesn't have declarations and it would be a pity having 143to introduce them just for this purpose. Using the explicit ``self.var`` solves 144this nicely. Similarly, for using instance variables, having to write 145``self.var`` means that references to unqualified names inside a method don't 146have to search the instance's directories. To put it another way, local 147variables and instance variables live in two different namespaces, and you need 148to tell Python which namespace to use. 149 150 151.. _why-can-t-i-use-an-assignment-in-an-expression: 152 153Why can't I use an assignment in an expression? 154----------------------------------------------- 155 156Starting in Python 3.8, you can! 157 158Assignment expressions using the walrus operator ``:=`` assign a variable in an 159expression:: 160 161 while chunk := fp.read(200): 162 print(chunk) 163 164See :pep:`572` for more information. 165 166 167 168Why does Python use methods for some functionality (e.g. list.index()) but functions for other (e.g. len(list))? 169---------------------------------------------------------------------------------------------------------------- 170 171As Guido said: 172 173 (a) For some operations, prefix notation just reads better than 174 postfix -- prefix (and infix!) operations have a long tradition in 175 mathematics which likes notations where the visuals help the 176 mathematician thinking about a problem. Compare the easy with which we 177 rewrite a formula like x*(a+b) into x*a + x*b to the clumsiness of 178 doing the same thing using a raw OO notation. 179 180 (b) When I read code that says len(x) I *know* that it is asking for 181 the length of something. This tells me two things: the result is an 182 integer, and the argument is some kind of container. To the contrary, 183 when I read x.len(), I have to already know that x is some kind of 184 container implementing an interface or inheriting from a class that 185 has a standard len(). Witness the confusion we occasionally have when 186 a class that is not implementing a mapping has a get() or keys() 187 method, or something that isn't a file has a write() method. 188 189 -- https://mail.python.org/pipermail/python-3000/2006-November/004643.html 190 191 192Why is join() a string method instead of a list or tuple method? 193---------------------------------------------------------------- 194 195Strings became much more like other standard types starting in Python 1.6, when 196methods were added which give the same functionality that has always been 197available using the functions of the string module. Most of these new methods 198have been widely accepted, but the one which appears to make some programmers 199feel uncomfortable is:: 200 201 ", ".join(['1', '2', '4', '8', '16']) 202 203which gives the result:: 204 205 "1, 2, 4, 8, 16" 206 207There are two common arguments against this usage. 208 209The first runs along the lines of: "It looks really ugly using a method of a 210string literal (string constant)", to which the answer is that it might, but a 211string literal is just a fixed value. If the methods are to be allowed on names 212bound to strings there is no logical reason to make them unavailable on 213literals. 214 215The second objection is typically cast as: "I am really telling a sequence to 216join its members together with a string constant". Sadly, you aren't. For some 217reason there seems to be much less difficulty with having :meth:`~str.split` as 218a string method, since in that case it is easy to see that :: 219 220 "1, 2, 4, 8, 16".split(", ") 221 222is an instruction to a string literal to return the substrings delimited by the 223given separator (or, by default, arbitrary runs of white space). 224 225:meth:`~str.join` is a string method because in using it you are telling the 226separator string to iterate over a sequence of strings and insert itself between 227adjacent elements. This method can be used with any argument which obeys the 228rules for sequence objects, including any new classes you might define yourself. 229Similar methods exist for bytes and bytearray objects. 230 231 232How fast are exceptions? 233------------------------ 234 235A try/except block is extremely efficient if no exceptions are raised. Actually 236catching an exception is expensive. In versions of Python prior to 2.0 it was 237common to use this idiom:: 238 239 try: 240 value = mydict[key] 241 except KeyError: 242 mydict[key] = getvalue(key) 243 value = mydict[key] 244 245This only made sense when you expected the dict to have the key almost all the 246time. If that wasn't the case, you coded it like this:: 247 248 if key in mydict: 249 value = mydict[key] 250 else: 251 value = mydict[key] = getvalue(key) 252 253For this specific case, you could also use ``value = dict.setdefault(key, 254getvalue(key))``, but only if the ``getvalue()`` call is cheap enough because it 255is evaluated in all cases. 256 257 258Why isn't there a switch or case statement in Python? 259----------------------------------------------------- 260 261You can do this easily enough with a sequence of ``if... elif... elif... else``. 262For literal values, or constants within a namespace, you can also use a 263``match ... case`` statement. 264 265For cases where you need to choose from a very large number of possibilities, 266you can create a dictionary mapping case values to functions to call. For 267example:: 268 269 functions = {'a': function_1, 270 'b': function_2, 271 'c': self.method_1} 272 273 func = functions[value] 274 func() 275 276For calling methods on objects, you can simplify yet further by using the 277:func:`getattr` built-in to retrieve methods with a particular name:: 278 279 class MyVisitor: 280 def visit_a(self): 281 ... 282 283 def dispatch(self, value): 284 method_name = 'visit_' + str(value) 285 method = getattr(self, method_name) 286 method() 287 288It's suggested that you use a prefix for the method names, such as ``visit_`` in 289this example. Without such a prefix, if values are coming from an untrusted 290source, an attacker would be able to call any method on your object. 291 292 293Can't you emulate threads in the interpreter instead of relying on an OS-specific thread implementation? 294-------------------------------------------------------------------------------------------------------- 295 296Answer 1: Unfortunately, the interpreter pushes at least one C stack frame for 297each Python stack frame. Also, extensions can call back into Python at almost 298random moments. Therefore, a complete threads implementation requires thread 299support for C. 300 301Answer 2: Fortunately, there is `Stackless Python <https://github.com/stackless-dev/stackless/wiki>`_, 302which has a completely redesigned interpreter loop that avoids the C stack. 303 304 305Why can't lambda expressions contain statements? 306------------------------------------------------ 307 308Python lambda expressions cannot contain statements because Python's syntactic 309framework can't handle statements nested inside expressions. However, in 310Python, this is not a serious problem. Unlike lambda forms in other languages, 311where they add functionality, Python lambdas are only a shorthand notation if 312you're too lazy to define a function. 313 314Functions are already first class objects in Python, and can be declared in a 315local scope. Therefore the only advantage of using a lambda instead of a 316locally defined function is that you don't need to invent a name for the 317function -- but that's just a local variable to which the function object (which 318is exactly the same type of object that a lambda expression yields) is assigned! 319 320 321Can Python be compiled to machine code, C or some other language? 322----------------------------------------------------------------- 323 324`Cython <https://cython.org/>`_ compiles a modified version of Python with 325optional annotations into C extensions. `Nuitka <https://www.nuitka.net/>`_ is 326an up-and-coming compiler of Python into C++ code, aiming to support the full 327Python language. 328 329 330How does Python manage memory? 331------------------------------ 332 333The details of Python memory management depend on the implementation. The 334standard implementation of Python, :term:`CPython`, uses reference counting to 335detect inaccessible objects, and another mechanism to collect reference cycles, 336periodically executing a cycle detection algorithm which looks for inaccessible 337cycles and deletes the objects involved. The :mod:`gc` module provides functions 338to perform a garbage collection, obtain debugging statistics, and tune the 339collector's parameters. 340 341Other implementations (such as `Jython <https://www.jython.org>`_ or 342`PyPy <https://www.pypy.org>`_), however, can rely on a different mechanism 343such as a full-blown garbage collector. This difference can cause some 344subtle porting problems if your Python code depends on the behavior of the 345reference counting implementation. 346 347In some Python implementations, the following code (which is fine in CPython) 348will probably run out of file descriptors:: 349 350 for file in very_long_list_of_files: 351 f = open(file) 352 c = f.read(1) 353 354Indeed, using CPython's reference counting and destructor scheme, each new 355assignment to *f* closes the previous file. With a traditional GC, however, 356those file objects will only get collected (and closed) at varying and possibly 357long intervals. 358 359If you want to write code that will work with any Python implementation, 360you should explicitly close the file or use the :keyword:`with` statement; 361this will work regardless of memory management scheme:: 362 363 for file in very_long_list_of_files: 364 with open(file) as f: 365 c = f.read(1) 366 367 368Why doesn't CPython use a more traditional garbage collection scheme? 369--------------------------------------------------------------------- 370 371For one thing, this is not a C standard feature and hence it's not portable. 372(Yes, we know about the Boehm GC library. It has bits of assembler code for 373*most* common platforms, not for all of them, and although it is mostly 374transparent, it isn't completely transparent; patches are required to get 375Python to work with it.) 376 377Traditional GC also becomes a problem when Python is embedded into other 378applications. While in a standalone Python it's fine to replace the standard 379malloc() and free() with versions provided by the GC library, an application 380embedding Python may want to have its *own* substitute for malloc() and free(), 381and may not want Python's. Right now, CPython works with anything that 382implements malloc() and free() properly. 383 384 385Why isn't all memory freed when CPython exits? 386---------------------------------------------- 387 388Objects referenced from the global namespaces of Python modules are not always 389deallocated when Python exits. This may happen if there are circular 390references. There are also certain bits of memory that are allocated by the C 391library that are impossible to free (e.g. a tool like Purify will complain about 392these). Python is, however, aggressive about cleaning up memory on exit and 393does try to destroy every single object. 394 395If you want to force Python to delete certain things on deallocation use the 396:mod:`atexit` module to run a function that will force those deletions. 397 398 399Why are there separate tuple and list data types? 400------------------------------------------------- 401 402Lists and tuples, while similar in many respects, are generally used in 403fundamentally different ways. Tuples can be thought of as being similar to 404Pascal records or C structs; they're small collections of related data which may 405be of different types which are operated on as a group. For example, a 406Cartesian coordinate is appropriately represented as a tuple of two or three 407numbers. 408 409Lists, on the other hand, are more like arrays in other languages. They tend to 410hold a varying number of objects all of which have the same type and which are 411operated on one-by-one. For example, ``os.listdir('.')`` returns a list of 412strings representing the files in the current directory. Functions which 413operate on this output would generally not break if you added another file or 414two to the directory. 415 416Tuples are immutable, meaning that once a tuple has been created, you can't 417replace any of its elements with a new value. Lists are mutable, meaning that 418you can always change a list's elements. Only immutable elements can be used as 419dictionary keys, and hence only tuples and not lists can be used as keys. 420 421 422How are lists implemented in CPython? 423------------------------------------- 424 425CPython's lists are really variable-length arrays, not Lisp-style linked lists. 426The implementation uses a contiguous array of references to other objects, and 427keeps a pointer to this array and the array's length in a list head structure. 428 429This makes indexing a list ``a[i]`` an operation whose cost is independent of 430the size of the list or the value of the index. 431 432When items are appended or inserted, the array of references is resized. Some 433cleverness is applied to improve the performance of appending items repeatedly; 434when the array must be grown, some extra space is allocated so the next few 435times don't require an actual resize. 436 437 438How are dictionaries implemented in CPython? 439-------------------------------------------- 440 441CPython's dictionaries are implemented as resizable hash tables. Compared to 442B-trees, this gives better performance for lookup (the most common operation by 443far) under most circumstances, and the implementation is simpler. 444 445Dictionaries work by computing a hash code for each key stored in the dictionary 446using the :func:`hash` built-in function. The hash code varies widely depending 447on the key and a per-process seed; for example, "Python" could hash to 448-539294296 while "python", a string that differs by a single bit, could hash 449to 1142331976. The hash code is then used to calculate a location in an 450internal array where the value will be stored. Assuming that you're storing 451keys that all have different hash values, this means that dictionaries take 452constant time -- O(1), in Big-O notation -- to retrieve a key. 453 454 455Why must dictionary keys be immutable? 456-------------------------------------- 457 458The hash table implementation of dictionaries uses a hash value calculated from 459the key value to find the key. If the key were a mutable object, its value 460could change, and thus its hash could also change. But since whoever changes 461the key object can't tell that it was being used as a dictionary key, it can't 462move the entry around in the dictionary. Then, when you try to look up the same 463object in the dictionary it won't be found because its hash value is different. 464If you tried to look up the old value it wouldn't be found either, because the 465value of the object found in that hash bin would be different. 466 467If you want a dictionary indexed with a list, simply convert the list to a tuple 468first; the function ``tuple(L)`` creates a tuple with the same entries as the 469list ``L``. Tuples are immutable and can therefore be used as dictionary keys. 470 471Some unacceptable solutions that have been proposed: 472 473- Hash lists by their address (object ID). This doesn't work because if you 474 construct a new list with the same value it won't be found; e.g.:: 475 476 mydict = {[1, 2]: '12'} 477 print(mydict[[1, 2]]) 478 479 would raise a :exc:`KeyError` exception because the id of the ``[1, 2]`` used in the 480 second line differs from that in the first line. In other words, dictionary 481 keys should be compared using ``==``, not using :keyword:`is`. 482 483- Make a copy when using a list as a key. This doesn't work because the list, 484 being a mutable object, could contain a reference to itself, and then the 485 copying code would run into an infinite loop. 486 487- Allow lists as keys but tell the user not to modify them. This would allow a 488 class of hard-to-track bugs in programs when you forgot or modified a list by 489 accident. It also invalidates an important invariant of dictionaries: every 490 value in ``d.keys()`` is usable as a key of the dictionary. 491 492- Mark lists as read-only once they are used as a dictionary key. The problem 493 is that it's not just the top-level object that could change its value; you 494 could use a tuple containing a list as a key. Entering anything as a key into 495 a dictionary would require marking all objects reachable from there as 496 read-only -- and again, self-referential objects could cause an infinite loop. 497 498There is a trick to get around this if you need to, but use it at your own risk: 499You can wrap a mutable structure inside a class instance which has both a 500:meth:`__eq__` and a :meth:`__hash__` method. You must then make sure that the 501hash value for all such wrapper objects that reside in a dictionary (or other 502hash based structure), remain fixed while the object is in the dictionary (or 503other structure). :: 504 505 class ListWrapper: 506 def __init__(self, the_list): 507 self.the_list = the_list 508 509 def __eq__(self, other): 510 return self.the_list == other.the_list 511 512 def __hash__(self): 513 l = self.the_list 514 result = 98767 - len(l)*555 515 for i, el in enumerate(l): 516 try: 517 result = result + (hash(el) % 9999999) * 1001 + i 518 except Exception: 519 result = (result % 7777777) + i * 333 520 return result 521 522Note that the hash computation is complicated by the possibility that some 523members of the list may be unhashable and also by the possibility of arithmetic 524overflow. 525 526Furthermore it must always be the case that if ``o1 == o2`` (ie ``o1.__eq__(o2) 527is True``) then ``hash(o1) == hash(o2)`` (ie, ``o1.__hash__() == o2.__hash__()``), 528regardless of whether the object is in a dictionary or not. If you fail to meet 529these restrictions dictionaries and other hash based structures will misbehave. 530 531In the case of ListWrapper, whenever the wrapper object is in a dictionary the 532wrapped list must not change to avoid anomalies. Don't do this unless you are 533prepared to think hard about the requirements and the consequences of not 534meeting them correctly. Consider yourself warned. 535 536 537Why doesn't list.sort() return the sorted list? 538----------------------------------------------- 539 540In situations where performance matters, making a copy of the list just to sort 541it would be wasteful. Therefore, :meth:`list.sort` sorts the list in place. In 542order to remind you of that fact, it does not return the sorted list. This way, 543you won't be fooled into accidentally overwriting a list when you need a sorted 544copy but also need to keep the unsorted version around. 545 546If you want to return a new list, use the built-in :func:`sorted` function 547instead. This function creates a new list from a provided iterable, sorts 548it and returns it. For example, here's how to iterate over the keys of a 549dictionary in sorted order:: 550 551 for key in sorted(mydict): 552 ... # do whatever with mydict[key]... 553 554 555How do you specify and enforce an interface spec in Python? 556----------------------------------------------------------- 557 558An interface specification for a module as provided by languages such as C++ and 559Java describes the prototypes for the methods and functions of the module. Many 560feel that compile-time enforcement of interface specifications helps in the 561construction of large programs. 562 563Python 2.6 adds an :mod:`abc` module that lets you define Abstract Base Classes 564(ABCs). You can then use :func:`isinstance` and :func:`issubclass` to check 565whether an instance or a class implements a particular ABC. The 566:mod:`collections.abc` module defines a set of useful ABCs such as 567:class:`~collections.abc.Iterable`, :class:`~collections.abc.Container`, and 568:class:`~collections.abc.MutableMapping`. 569 570For Python, many of the advantages of interface specifications can be obtained 571by an appropriate test discipline for components. 572 573A good test suite for a module can both provide a regression test and serve as a 574module interface specification and a set of examples. Many Python modules can 575be run as a script to provide a simple "self test." Even modules which use 576complex external interfaces can often be tested in isolation using trivial 577"stub" emulations of the external interface. The :mod:`doctest` and 578:mod:`unittest` modules or third-party test frameworks can be used to construct 579exhaustive test suites that exercise every line of code in a module. 580 581An appropriate testing discipline can help build large complex applications in 582Python as well as having interface specifications would. In fact, it can be 583better because an interface specification cannot test certain properties of a 584program. For example, the :meth:`append` method is expected to add new elements 585to the end of some internal list; an interface specification cannot test that 586your :meth:`append` implementation will actually do this correctly, but it's 587trivial to check this property in a test suite. 588 589Writing test suites is very helpful, and you might want to design your code to 590make it easily tested. One increasingly popular technique, test-driven 591development, calls for writing parts of the test suite first, before you write 592any of the actual code. Of course Python allows you to be sloppy and not write 593test cases at all. 594 595 596Why is there no goto? 597--------------------- 598 599In the 1970s people realized that unrestricted goto could lead 600to messy "spaghetti" code that was hard to understand and revise. 601In a high-level language, it is also unneeded as long as there 602are ways to branch (in Python, with ``if`` statements and ``or``, 603``and``, and ``if-else`` expressions) and loop (with ``while`` 604and ``for`` statements, possibly containing ``continue`` and ``break``). 605 606One can also use exceptions to provide a "structured goto" 607that works even across 608function calls. Many feel that exceptions can conveniently emulate all 609reasonable uses of the "go" or "goto" constructs of C, Fortran, and other 610languages. For example:: 611 612 class label(Exception): pass # declare a label 613 614 try: 615 ... 616 if condition: raise label() # goto label 617 ... 618 except label: # where to goto 619 pass 620 ... 621 622This doesn't allow you to jump into the middle of a loop, but that's usually 623considered an abuse of goto anyway. Use sparingly. 624 625 626Why can't raw strings (r-strings) end with a backslash? 627------------------------------------------------------- 628 629More precisely, they can't end with an odd number of backslashes: the unpaired 630backslash at the end escapes the closing quote character, leaving an 631unterminated string. 632 633Raw strings were designed to ease creating input for processors (chiefly regular 634expression engines) that want to do their own backslash escape processing. Such 635processors consider an unmatched trailing backslash to be an error anyway, so 636raw strings disallow that. In return, they allow you to pass on the string 637quote character by escaping it with a backslash. These rules work well when 638r-strings are used for their intended purpose. 639 640If you're trying to build Windows pathnames, note that all Windows system calls 641accept forward slashes too:: 642 643 f = open("/mydir/file.txt") # works fine! 644 645If you're trying to build a pathname for a DOS command, try e.g. one of :: 646 647 dir = r"\this\is\my\dos\dir" "\\" 648 dir = r"\this\is\my\dos\dir\ "[:-1] 649 dir = "\\this\\is\\my\\dos\\dir\\" 650 651 652Why doesn't Python have a "with" statement for attribute assignments? 653--------------------------------------------------------------------- 654 655Python has a 'with' statement that wraps the execution of a block, calling code 656on the entrance and exit from the block. Some languages have a construct that 657looks like this:: 658 659 with obj: 660 a = 1 # equivalent to obj.a = 1 661 total = total + 1 # obj.total = obj.total + 1 662 663In Python, such a construct would be ambiguous. 664 665Other languages, such as Object Pascal, Delphi, and C++, use static types, so 666it's possible to know, in an unambiguous way, what member is being assigned 667to. This is the main point of static typing -- the compiler *always* knows the 668scope of every variable at compile time. 669 670Python uses dynamic types. It is impossible to know in advance which attribute 671will be referenced at runtime. Member attributes may be added or removed from 672objects on the fly. This makes it impossible to know, from a simple reading, 673what attribute is being referenced: a local one, a global one, or a member 674attribute? 675 676For instance, take the following incomplete snippet:: 677 678 def foo(a): 679 with a: 680 print(x) 681 682The snippet assumes that "a" must have a member attribute called "x". However, 683there is nothing in Python that tells the interpreter this. What should happen 684if "a" is, let us say, an integer? If there is a global variable named "x", 685will it be used inside the with block? As you see, the dynamic nature of Python 686makes such choices much harder. 687 688The primary benefit of "with" and similar language features (reduction of code 689volume) can, however, easily be achieved in Python by assignment. Instead of:: 690 691 function(args).mydict[index][index].a = 21 692 function(args).mydict[index][index].b = 42 693 function(args).mydict[index][index].c = 63 694 695write this:: 696 697 ref = function(args).mydict[index][index] 698 ref.a = 21 699 ref.b = 42 700 ref.c = 63 701 702This also has the side-effect of increasing execution speed because name 703bindings are resolved at run-time in Python, and the second version only needs 704to perform the resolution once. 705 706 707Why don't generators support the with statement? 708------------------------------------------------ 709 710For technical reasons, a generator used directly as a context manager 711would not work correctly. When, as is most common, a generator is used as 712an iterator run to completion, no closing is needed. When it is, wrap 713it as "contextlib.closing(generator)" in the 'with' statement. 714 715 716Why are colons required for the if/while/def/class statements? 717-------------------------------------------------------------- 718 719The colon is required primarily to enhance readability (one of the results of 720the experimental ABC language). Consider this:: 721 722 if a == b 723 print(a) 724 725versus :: 726 727 if a == b: 728 print(a) 729 730Notice how the second one is slightly easier to read. Notice further how a 731colon sets off the example in this FAQ answer; it's a standard usage in English. 732 733Another minor reason is that the colon makes it easier for editors with syntax 734highlighting; they can look for colons to decide when indentation needs to be 735increased instead of having to do a more elaborate parsing of the program text. 736 737 738Why does Python allow commas at the end of lists and tuples? 739------------------------------------------------------------ 740 741Python lets you add a trailing comma at the end of lists, tuples, and 742dictionaries:: 743 744 [1, 2, 3,] 745 ('a', 'b', 'c',) 746 d = { 747 "A": [1, 5], 748 "B": [6, 7], # last trailing comma is optional but good style 749 } 750 751 752There are several reasons to allow this. 753 754When you have a literal value for a list, tuple, or dictionary spread across 755multiple lines, it's easier to add more elements because you don't have to 756remember to add a comma to the previous line. The lines can also be reordered 757without creating a syntax error. 758 759Accidentally omitting the comma can lead to errors that are hard to diagnose. 760For example:: 761 762 x = [ 763 "fee", 764 "fie" 765 "foo", 766 "fum" 767 ] 768 769This list looks like it has four elements, but it actually contains three: 770"fee", "fiefoo" and "fum". Always adding the comma avoids this source of error. 771 772Allowing the trailing comma may also make programmatic code generation easier. 773