Talk:Storage class

other languages

Seems that this currently redirects only to C's storage class.

There is currently discussion for PL/I's storage class, which mentions Fortrans named COMMON. Some ideas, such as static vs. automatic, apply to many languages. There is also internal vs. external. Gah4 (talk) 21:56, 7 August 2024 (UTC)[reply]

Yeah, this should be a page for the concept, which dates back, under the name "storage class", at least to PL/I. Guy Harris (talk) 22:37, 7 August 2024 (UTC)[reply]

I hadn't actually thought back to where it came from. Fortran only had static storage until Fortran 90. If forget now if ALGOL had static storage, along with automatic storage needed for recursion. It does seem that the names originated in PL/I. Gah4 (talk) 23:57, 7 August 2024 (UTC)[reply]

If forget now if ALGOL had static storage Section 5 "Declarations" of this version of the Algol 60 report says

A declaration my be marked with the additional declarator own. This has the following effect: upon a reentry into the block, the values of own quantities will be unchanged from their values at the last exit, while the values of declared variables which are not marked as own are undefined.

which sounds as if it might be the equivalent of a local static variable in C. Guy Harris (talk) 00:14, 8 August 2024 (UTC)[reply]

There is interest in making this a more general article on storage class, likely with mentions of, and links to, pages about other languages. PL/I seems to be the origin of some of the discussion and naming of storage class, among others with the names borrowed for C, except for CONTROLLED. Some languages leave storage class more ambiguous. Gah4 (talk) 21:11, 23 March 2025 (UTC)[reply]

draft

I have been working on editing this article in draft:Storage class, adding other languages and history. Peter Flass (talk) 20:39, 25 March 2025 (UTC)[reply]

Notability

I presume Notability is not a problem, but do we have actual sources to indicate the notability? Seems to me that many consider it so obvious, that they don't mention any reason for it to be notable.

As I noted, it seems to me that PL/I is where it got better known. Fortran pretty much ignored it, and ALGOL barely covered it. I don't know COBOL at all, and as far as I know, does (or did) only do static allocation. (Does COBOL allow recursion yet?)

Might be nice to have some sources actually discuss its notability, that is, usefulness. Gah4 (talk) 18:42, 30 March 2025 (UTC)[reply]

Does COBOL allow recursion yet? To quote ISO/IEC IS 1989:2001 "Information technology — Programming languages, their environments and system software interfaces — Programming language COBOL":

8.6.5 Common, initial, and recursive attributes

A program can be described with attributes that affect its initial state or that define the manner in which it can be called.

A common program is one that is directly contained within another program and that can be called by programs directly or indirectly contained in that other program, as described in 8.4.5.2, Scope of program-names. The common attribute is attained by specifying the COMMON clause in a program's identification division. When the COMMON clause is not specified, a contained program that is not recursive may be called only from the directly- containing program. The COMMON clause facilitates the writing of subprograms that can be used by all the programs contained within a program.

An initial program is one whose program state is initialized when the program is called. During the process of initializing an initial program, that program's internal data is initialized as described in 14.5.2, State of a function, method, object, or program. The initial attribute is attained by specifying the INITIAL clause in the program's identification division.

A recursive program may call itself directly or indirectly. The program's internal data is initialized as described in 14.5.2, State of a function, method, object, or program. The recursive attribute is attained by specifying the RECURSIVE clause in the program’s identification division.

Functions and methods are always recursive. Their data is initialized in the same way as recursive programs.

If neither the INITIAL nor RECURSIVE clause is specified in a program's identification division, the program's data is in the last-used state on other than the first activation of the program as described in 14.5.2, State of a function, method, object, or program. The program cannot be activated while it is active unless RECURSIVE is specified.

so it's supported recursion since at least what was, according to COBOL § COBOL 2002 and object-oriented COBOL, apparently called "COBOL 2002". (COBOL § Object-oriented programming indicates that COBOL 2002 ADDED 1 to COBOL.)

As for storage classes, to quote E.4.2 Recursive and initial programs:

Early COBOL programs always had data in its last-used state and did not allow calling a program when the program was active. Initial and recursive programs allow data to be initialized on every invocation and recursive programs allow programs to be called when they are active. Functions and methods are always recursive.

An initial program is one in which working-storage section internal data items and internal file connectors are set to their initial state whenever the program is called. Data items and file connectors declared as external are left in their last-used state. When the program exits (with EXIT PROGRAM or GOBACK), all programs that are contained in it are cancelled (as if CANCEL were executed for each one) and any internal file connectors are closed. This type of program is most often used when it is not necessary to retain data from one invocation to the next. To retain specific data, external data items or file connectors can be used. In many implementations of initial programs, working-storage is allocated "on the stack", which can conserve space because it goes away when the program terminates.

A recursive function, method, or program is one that can be called when it is already active. For example, program A can call B that can in turn call A again. Or, A can call A as illustrated in the example in E.4.1.3 where the function factorial calls itself. Typically, in a recursive program a local-storage section is specified for data that is initialized on each invocation of the program (this data is called automatic data). Working-storage data is static and is therefore in its last-used state on every invocation. The programmer should be aware of this because it can cause unexpected results. For example, if you had a counter "xyz" in the working-storage in one recursion of a program and you added to it, when you got back to another recursion it would be one more than it was before. You might or might not want this to happen.

so there are automatic and static storage classes. Guy Harris (talk) 19:17, 30 March 2025 (UTC)[reply]

Per WP:GNG A topic is presumed to be suitable for a stand-alone article or list when it has received significant coverage in reliable sources that are independent of the subject.. In addition per WP:BURDEN All content must be verifiable. The burden to demonstrate verifiability lies with the editor who adds or restores material, and it is satisfied by providing an inline citation to a reliable source that directly supports the contribution. The article needs to show the subject is notable and also references need to be added for most of the content. --John B123 (talk) 19:33, 30 March 2025 (UTC)[reply]

It should be compiler books that actually discuss the notability, but I don't have any of those nearby now. Gah4 (talk) 13:44, 31 March 2025 (UTC)[reply]

Or books and papers that teach the concepts of programming languages, plural (rather than texts for a particular programming language), or of programming language design. Guy Harris (talk) 20:59, 31 March 2025 (UTC)[reply]

independent

Is storage class really (and always) independent of type and value? It should be mostly independent, but I am suspecting that there are exceptions. C used to not allow initializers for auto arrays. Maybe that doesn't count, though. PL/I has ENTRY variables and LABEL variables, which might have some restrictions on them. Gah4 (talk) 18:15, 3 April 2025 (UTC)[reply]

ENTRY

What do you call the values of PL/I ENTRY variables? I think IBM calls them EXTERNAL, though others might call them GLOBAL. I can't find a WP page describing them, which is the reason to ask. It seems that there is Function pointer with redirects from Subroutine pointer and Procedure Pointer. Maybe that isn't so far off. Still doesn't seem quite right, though. Gah4 (talk) 18:38, 3 April 2025 (UTC)[reply]

I think ENTRY variables can be EXTERNAL or INTERNAL, independent of the contents. I think the values can be either, too. Basically the variable will have the address and, for internal procedures, a pointer to the stack frame for the containing procedure, so the code can access variables from higher-level procs. At least this is what Iron Spring PL/I has. This is the minimum, so IBM PL/I may hsve the same or more (I haven’t looked). Peter Flass (talk) 00:58, 4 April 2025 (UTC)[reply]

That is supposed to be the answer to the above question, but otherwise ... for this one, I am wondering what you call such values. IBM tends to call them external symbols, that go in the external symbol directory. So, I would say that an ENTRY variable has a value that is an external symbol. But I was trying to find the more general, or CS, name for such. It seems that the variables are function pointers in C, or maybe procedure pointers in Fortran. It doesn't sound right to say that the value is a function or procedure, though. And since PL/I has the ENTRY statement, the variable can reference on of those. There is a page for Function pointer and redirect for Procedure pointer, but I wanted to find the page for the value of such. It should have a page ... Gah4 (talk) 07:00, 4 April 2025 (UTC)[reply]

The value of an ENTRY variable is an ENTRY reference, which could be either a PROCEDURE or an ENTRY statement. I’m not sure it deserves a page of its own, but maybe a section in the Procedure pointer article. Although the current IBM PL/I manual mentions ENTRY variables, I couldn’t find a discussion, so I’ll have to check other references, but I believe the value can be either an external or internal entry. Use for internal entries is why it needs context in addition to a bare address of the entry. C functions, of course, doesn’t have any context. I think any reference to the ESD is TMI for discussion of ENTRY variables. Peter Flass (talk) 12:59, 4 April 2025 (UTC)[reply]

Oh, yes. The ESD part was to be sure to explain what I was asking. And yes, as well as I know, PL/I does allow internal procedures. I was trying to explain it, and that was the first one I thought about. But okay, it isn't quite external name or ESD, as that also applies to external data names. Gah4 (talk) 03:25, 5 April 2025 (UTC)[reply]