Talk:IBM 3270

On the Read Modified
On the Read Modified technique: Modern users however, sometimes find this system extremely bizarre, since it is very different to any user interface encountered in the consumer market. Isn't this just the form-submission paradigm every web user is intimately familiar with? grendel|khan 18:19, 2005 May 4 (UTC)


 * Yes, exactly. In many ways the 3270 worked very similarly to the way we do forms with a web browser: a screen of text was sent to the terminal, the user interacted with the text in some fashion, and a special key on the terminal was used to submit the page back to the mainframe. Actually, in the mid-1980s this seemed stranger than it does now--at least to people used to working with Unix, because all terminal interactions were character-I/O oriented.


 * It's also easy to see how this process cleanly integrated into a batch processing mainframe environment, because each submission was a single batch operation. (Editing text files was a bit funny, because you could only see one page at a time and there was no line-by-line scrolling.... at least on Wylbur.)--12.103.251.203 22:25, 20 September 2005 (UTC)


 * I added a description of attribute bits, and included a sentence on this technique. Yes - the first time I how the web browsers were operating it took me back to my CICS days. Peter Flass (talk) 12:56, 8 July 2012 (UTC)

LU's? What's that?
I often read about LUs in the TN3270 context. What are LUs? Perhaps that means Logical Unit? But what does that mean really in this context? — Preceding unsigned comment added by 65.245.11.134 (talk) 21:24, 23 June 2006 (UTC)


 * Yes, Logical Units. In the Mainframe side, Communications Server (VTAM) must have defined every unit that can connect to the system. LUs and PUs (yes, Physical Units) are SNA concepts. — Preceding unsigned comment added by 195.76.110.50 (talk) 02:37, 26 August 2006 (UTC)

Unfair external link removal
Has no sense to remove an useful external link like the one i added when other links of the competence can be present in the list. It is not only unfair, it is more than that. If you are going to remove a link to a product, remove them all, which is the difference Femto? In fact, i saw that one of the people that had removed some links is a kwnon member of one of the competence companies. I repeat, has no sense to remove ONE link, or you let me add my link, or you remove all external links, but be fair. — Preceding unsigned comment added by 200.61.172.5 (talk) 15:31, 17 May 2007 (UTC)


 * Next time, don't try to add the link to 9 different articles in 10 minutes and maybe you won't be accused of spamming. RossPatterson 00:14, 18 May 2007 (UTC)


 * Your concerns are covered at WP:EL and WP:SPAM.-- I already forgot  talk  22:36, 18 May 2007 (UTC)

3277 GA information is incorrect
The 3277 GA (Graphics Attachment) was an add-in card for a 3277 that drove a Tektronix 4015 directly in vector mode, not via RS232. The resolution of the graphics device was 4096x4096, not 1024x768. In fact, the resolution was so high that the GA could be used to display 512x512 gray-scale images by stroking a special "font" that displayed each "pixel" of the image as a character. Light pixels had lots of strokes, dark pixels had relatively few strokes. Zen3d (talk) 00:54, 15 January 2009 (UTC)


 * Actually it was bit more than "an add-in card", it was whole gate full. A new gate was installed on the other side of the CRT neck from the regular gate.  I know, I converted a regular 3277 to a 3277 GA once.  Terry (Ex-IBM CE). — Preceding unsigned comment added by 86.136.24.229 (talk) 14:45, 21 June 2011 (UTC)


 * The resolution of the Tektronix screen was actually 4k x 3k with the added ability to display limited dynamic vector graphics (in orange) on top of the green static image. The 3277-GA workstation was used extensively to design the 3090 series of IBM mainframes. We were able to project a full page of a chip logic diagram on the screen or an image of a 100 chip TCM and its connections. The "rubber band mode" was used to visualize inter-chip connections as part of a placement analysis tool. — Preceding unsigned comment added by 65.208.210.98 (talk) 22:48, 28 June 2016 (UTC)


 * The Tektronix 40xx devices use a storage CRT, which has no resolution. It is an analog storage medium. The beam control hardware has some DACs with some number of bits, but if one wanted to, one could replace all that logic.  (I don't know exactly how IBM did it.)  Gah4 (talk) 23:39, 30 September 2018 (UTC)

Picture should be of the real thing
A picture of a real terminal would be better than a screenshot of an emulator. —Preceding unsigned comment added by Ljosa (talk • contribs) 17:41, 13 April 2009 (UTC)


 * Added a 3279 photo. Will try to shoot a 3270 with stock keyboard next month. Shieldforyoureyes (talk) 14:34, 21 July 2009 (UTC)


 * That picture isn't a picture of a (regular) 3279.  It's a special version of a 3279 that is used to (among other things) IML the system, a lamp test, stop/start, and other things, and has lights (lamps) to indicate the system's status.     -- GerardSchildberger (talk) 04:57, 14 January 2020 (UTC)

I just (Oct., 2010) renamed that picture because it isn't a 3279, it is a console. (It has Stop Start keys) Regards, Terry — Preceding unsigned comment added by 86.134.48.193 (talk) 18:22, 11 October 2010 (UTC)


 * I'd like to find a picture of an "original" 3270 (3275 or 3277). Peter Flass (talk) 12:58, 8 July 2012 (UTC)


 * Seems to me that emulation is common enough that one should have pictures of both. Gah4 (talk) 23:40, 30 September 2018 (UTC)


 * How is an "IBM 3279 Model 2C Color Display Console" not a 3279? Shmuel (Seymour J.) Metz Username:Chatul (talk) 19:06, 14 January 2020 (UTC)


 * When it is a console, not a terminal.  A "console" is used for more than a terminal, it is used to control a computer (or a PC).     -- GerardSchildberger (talk) 19:51, 14 January 2020 (UTC)


 * Whether it is a console, a terminal or a pink elephant, it is still a 3279 and referred to by IBM as such. Shmuel (Seymour J.) Metz Username:Chatul (talk) 19:52, 15 January 2020 (UTC)

Brazilian manufacture
I have restored that contribution, as it appears to be both pertinent and, according to anecdotal evidence, correct.

--ClickRick (talk) 09:23, 28 May 2009 (UTC)

Physical description
When I added the 3277 photo I wanted to include the dimensions, and especially the weight, which I thought would be a surprise to people today. Unfortunately I couldn't find a manual or announcement that has this information. Bitsavers only has the 3278. If anyone has the info, please include it. Peter Flass (talk) 12:36, 9 August 2012 (UTC)

Data Stream Example
I added a short example of a data stream sent to the 3270. Unfortunately I no longer have access to a 3270 to verify it, Likewise I'd like to include what the terminal would send back to the host on a read, but I'd rather just not wing it. If anyone could verify this example, or substitute your own, it would be nice. Peter Flass (talk) 00:12, 10 August 2012 (UTC)


 * The example is correct for a 3277-2 if there are not already fields in row 24. In the absence of information about what was in row 24, you need to use, e.g., an RA sequence, to clear the remainder of row 24. Clearing through column 79 will leave you at column 80, so you won't need the SBA before the SF/F0.


 * Of course, for a 3278-5 the buffer addresses are wrong.


 * As for the results of a read, it depends. What type of read, and what triggered it?
 * short read?
 * read modified?
 * read buffer?


 * Meanwhile, has lots of information.


 * I don't have access to a physical 3270 either, but this stuff has been bread and butter for decades, so if you give me the scenario you want I should be able to tell you the read results. Shmuel (Seymour J.) Metz Username:Chatul (talk) 21:57, 1 July 2016 (UTC)

3472 Infowindow
Added a placeholder for this. Anyone able to beef it up a little? Such as the announce date? You'd think I'd remember. :-) Martin Packer (talk) 17:02, 20 July 2013 (UTC)

mixes ASCII and EBCDIC
The table in IBM 3270 has rows and columns correctly labeled for EBCDIC, but the hexadecimal values shown are for ASCII rather than for EBCDIC. Also, neither of them is a 16 bit character set, and the values should be two hexadecimal digits rather than four. Shmuel (Seymour J.) Metz Username:Chatul (talk) 16:03, 17 September 2018 (UTC)


 * Those numbers are the matching Unicode code points, as used in many of the tables. I did try to remove these (putting them into the tooltips) as I thought they are ugly and confusing, but it was reverted, so I gave up.Spitzak (talk) 19:01, 17 September 2018 (UTC)


 * Should there also be a table for 3270 ASCII encoding? Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:37, 18 September 2018 (UTC)

Control characters
The 3270 controllers are available in BSC (BiSynch), SNA, local non-SNA and local SNA versions. Some of the control characters are only relevant to BSC:


 * ACK 0 ('1070'x)
 * ACK 1 ('1061')
 * DLE
 * EOT
 * ESC
 * ETX
 * ITB
 * NAK
 * RVI ('1070')
 * SOH
 * STX
 * SYN
 * TTD (STX ENQ)
 * WACK ('106B'x)

The relevant reference for BSC 3270 controllers is http://bitsavers.org/pdf/ibm/3270/GA23-0060-0_3271_3272_3275_Description_and_Programmers_Guide_Nov80.pdf, with equivalents for newer controllers.

Some of the control characters are specific to the 3270 and have no Unicode equivalents:


 * DUP
 * IC
 * EUA
 * FM
 * GE
 * MF
 * PT
 * RA
 * SA
 * SBA
 * SF
 * SFE

The relevant manual for control characters unique to the 3270 is http://bitsavers.org/pdf/ibm/3270/GA23-0059-4_3270_Data_Stream_Programmers_Reference_Dec88.pdf or a later edition. Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:22, 18 September 2018 (UTC)

Of what Earthly use are decimal code points for 3270 control characters?
User:Matthiaspaul claimed that table 1 of IBM 3270 was useless without the decimal numbers and reverted an edit by User:Spitzak; the table is perfectly useful without the decimal numbers and, in fact, the decimal numbers are useless; anybody programming for the 3270 uses hexadecimal or symbolic references to control characters. Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:42, 25 September 2018 (UTC)

ASCII code points
I think having the ASCII values in the character set table is more than confusing. I haven"t been following the various changes and undos, butI think mixing ASCII and EBCRIC in one table is a bad idea(tm). There is a 3270 ASCII character set, and I think if people want to see ASCII there should be a separate ASCII table. Peter Flass (talk) 19:18, 26 September 2018 (UTC)


 * Do you mean the unicode code points? They do unambiguously identify the consensus as to what each character is. However I agree they really serve no purpose being shown in the table except to provide a source for somebody writing a lookup table to translate this set, and would prefer they be in a tooltip or even in comments in the wiki. Unfortunately all my attempts to remove these (and the decimal indexes, and even *octal* indexes) cluttering these tables have been reverted by User:Matthiaspaul on the claim that there is "no consensus" for the edits. I would vastly prefer a highly simplified table that highlights the characters and is small enough so that you can actually look at more than one at the same time, an example of my ideal layout is in talk:ISO/IEC_8859-1.


 * If you can speak up and ask others to so as to indicate anything that might be considered a "consensus".

Spitzak (talk) 19:46, 26 September 2018 (UTC)


 * What consensus is there that, e.g., a 3270 DUP is a Unicode File Separator? Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:08, 28 September 2018 (UTC)


 * There is consensus for a specific consistent format for character set tables all over Wikipedia (a format currently forcefully overridden in dozens of articles by Spitzak's edit-warring). This format includes the decimal index for easier reference. There is also consensus to show the Unicode glyph and code, but this should obviously happen only where a Unicode equivalent exists. Therefore, the Unicode should be removed from DUP.
 * --Matthiaspaul (talk) 21:49, 29 September 2018 (UTC)

I was more concerned with the 4-digit hex values. It wasn't obvious to me that they were Unicode. Maybe put in the U+. Chatul is right that a lot of the orders (not controls) have no Unicode squivalents. Peter Flass (talk) 19:17, 28 September 2018 (UTC)


 * Yea I put in a citation-needed at the top about the control codes, I have my doubts the Unicode translations make any sense.


 * Would like any opinions about whether the decimal indexes and/or unicode numbers should be visible. IMHO it would be much better if this table just had character glyphs.

Spitzak (talk) 19:44, 28 September 2018 (UTC)


 * Per long established consensus and prior discussions the decimal indexes should be visible because the table's utility value is significantly reduced without this information. You cannot expect readers to use a formula to calculate the decimal index value - Wikipedia is not for techies only, but should be accessible to anyone. If there were no space constraints, we should ideally also list the hex index alongside the decimal index, however, the fact that the tables are arranged with 16 characters per line and that hex values are typically used by computer experts, makes the "calculation" of hex values easy enough to make it acceptable to list them only in the top line and left column (but not the other way around).
 * However, Unicode (or ASCII) equivalent codes should be given only where an equivalent exists. Here, this includes the letters, digits and punctuation characters but not some of the control codes.
 * --Matthiaspaul (talk) 21:49, 29 September 2018 (UTC)


 * Who would use the decimal code points for 3270 orders and for EBCDIC code points, and what would they use the number as an index into? Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:39, 30 September 2018 (UTC)
 * Anyone who is not familiar with hex values - that is the vast majority of people on this planet... One of our fundamental goals is to provide information in a way which remains accessible to laymen.
 * What are laymen going to do with the table? I agree that the article should be understandable to someone who doesn't know hex, but in this case they are labels for code points. I suppose decimal has some some use in an ASCII table, but much less here.  Since the default base for IBM assemblers is decimal, I suppose one could code a series of characters in an AL1 constant, but why? Gah4 (talk) 23:48, 30 September 2018 (UTC)
 * The index number represents the code point, a fundamental property of any character set.
 * --Matthiaspaul (talk) 20:26, 30 September 2018 (UTC)
 * The question is not why to have an index, but rather why to have an index in decimal when all of the EBCDIC documentation shows code points only in hexadecimal. A layman who is not familiar with hexadecimal will not be able to use the decimal index, so why put it there? Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:53, 1 October 2018 (UTC)


 * I am aware of EBCDIC documentation showing code points in decimal and hex.
 * So apparently you have not been able to find any documentation that does not use hex?Spitzak (talk) 20:12, 15 October 2018 (UTC)
 * The index is helpful when comparing character sets side by side.
 * Making the tables smaller is by far the most useful thing to compare them.Spitzak (talk) 20:12, 15 October 2018 (UTC)
 * We have an established standard format (including the decimal index) used in almost all (that is hundreds of) character set tables all over Wikipedia - I can't see a pressing reason why we would have to deviate from this format in this article. Different formats only make comparisons more difficult.
 * --Matthiaspaul (talk) 12:00, 3 October 2018 (UTC)


 * While Unicode DC1-DC4 are the equivalents of EBCDIC DC1-DC4, I question whether it is appropriate to show BSC control characters (DLE) and 3270 orders (SBA, EUA, IC) using the code points for DC1-DC4 as equivalent to the general Unicode DC1-DC4. Shmuel (Seymour J.) Metz Username:Chatul (talk)
 * I agree with you here. That's why I added question marks to a few Unicode codes I found questionable. IMHO, those codes should be removed from the table. But that's just fixing bugs in the table not questioning the table as a whole...
 * --Matthiaspaul (talk) 20:26, 30 September 2018 (UTC)

Let's return to the main subject. We are not talking about the Unicode for a few of the control characters. We are talking about the decimal cell indexes you insist are "vital" and a "consensus" but that at least 3 people have said they don't want. I believe it is totally wrong to put information about the cell's location into the cell, such information should be in the row/column headers. I also think it is incredibly ugly and would prefer if Wikipedia did not look like crap!Spitzak (talk) 20:12, 15 October 2018 (UTC)

Inappropriate reference - VM/SP Multipoint for 3270 adds support to CP, not to CMS
IBM 3270 gives an inappropriate reference in Conversational Monitor System (CMS) in VM/SP has support for the 3270.

The article Remote 3270s Gain Access to VM Hosts has nothing to do with CMS support for the 3270, which existed in the free Virtual Machine Facility/370 long before VM/SP. In fact, support for remote 3270s was already in VM/370 R2. What VM/SP Multipoint for 3270 added was support for multidrop (polled) 3270s and does not change CMS. I would probably have given a reference for the earlier VM/370 in place of or in addition to a reference for VM/SP, but the existing reference is irrelevant. Shmuel (Seymour J.) Metz Username:Chatul (talk) 15:54, 15 March 2019 (UTC)


 * I think VM/SP added support for full-screen 3270s, which made them a lot more useful. Peter Flass (talk) 17:25, 15 March 2019 (UTC)


 * Full screen support was there in VM/370. VM/SP added XEDIT and later VSCREEN. Shmuel (Seymour J.) Metz Username:Chatul (talk) 19:38, 15 March 2019 (UTC)

models
The article does not mention the 3180 or 3192, although one of the references covers the 3192G. Does anybody have the dates for them? Shmuel (Seymour J.) Metz Username:Chatul (talk) 19:53, 16 September 2019 (UTC)


 * google> "3180" "1984" "terminal" found 1984 - IBM's announcement date is March 20, 1984 for "IBM 3180 DISPLAY STATION MODELS 1 AND 2" Pi314m (talk) 05:54, 2 October 2019 (UTC)

3270 graphics and organization of the article
After adding information on the color modes supported on the 3279, I realized that those data really belong in a central location rather than being replicated in the description of every relevant model. Further, I believe that current references to graphics are misleading. I'm soliciting advice on what goes where.

First, the 3270 offers three very different types of graphics support


 * 1) The 3277 Graphics Attachment has an attached Tektronix CRT and is incompatible with all other 3270 graphics.
 * 2) 3270 models with the Programmed symbol sets  (PSS) feature are supported by Graphical Data Display Manager (GDDM) and by the Presentation Graphics Feature (PGF) of GDDM.
 * 3) 3270 models with the All Point Addressable (APA) feature are supported by GDDM and by the Presentation Graphics Feature (PGF) of GDDM.

Typically models with a G in the number support APA, but GDDM and GDDM-PGF support some models without APA.

Second, the 3270 has two types of color support briefly described in IBM 3270.


 * 1) Basic color (4 implicit colors)
 * 2) Extended color  (7 explicit colors) Shmuel (Seymour J.) Metz Username:Chatul (talk) 22:23, 16 January 2020 (UTC)

WWW browsers are not thin clients
It is likely that Tim Berners-Lee thought that WWW browsers should be thin clients, and I wish that they had remained so, but Todays web is dominated by sites using client-side scripts. Shmuel (Seymour J.) Metz Username:Chatul (talk) 01:43, 21 September 2020 (UTC)

External links format
I'd like to change the format of IBM 3270 to


 * Use cite IETF
 * Use cite manual
 * Use cite web
 * Use 3270 data stream manual from bitsavers instead of newer versions with dead links
 * Add ref where appropriate
 * use a definition list for most references

Following that I would like to add sfnreferences with loc, p and pp as appropriate, e.g., The IBM 3270 is a family of block oriented, using the 3270 data stream., it also supports partitioning. with these external links


 * noting the unveiling of the 3270 display system in 1971
 * rbanffy/3270font: A TTF remake of the font from the 3270
 * 3270Intro
 * 3270DS
 * RFC1041
 * RFC1576
 * RFC2355
 * RFC1041
 * RFC1576
 * RFC2355
 * RFC1576
 * RFC2355
 * RFC2355


 * thank you. We're not using Help:Shortened footnotes in the article but I don't see a problem introducing some. Converting external links to references is always an improvement as far as I'm concerned. ~Kvng (talk) 13:56, 10 November 2020 (UTC)


 * To clarify, I'm not proposing to remove the External Links section, although I am not averse to doing so if there are no objections. What I was proposing was to add document to external links in definition list format, with the tags being the values to use in sfn.


 * As an alternative, I could include the document as or give the citation inside the  as


 * tag

Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:51, 10 November 2020 (UTC)

Remove home-grown controller or add commercially available controllers
IBM 3270 should either be removed as not notable or be greatly expanded to show all of the many OEM controllers that were successfully marketed before the home-grown controller currently described in that section, e.g., 1371/1372 Memorex 1372, newer Memorex controllers listed in http://bitsavers.org/pdf/memorex/terminal/Memorex_3270_Compatible_Terminals_Jan86.pdf. Shmuel (Seymour J.) Metz Username:Chatul (talk) 03:30, 16 May 2021 (UTC)


 * I don't disagree about possible expansion. I think the controller listed is interesting because it's built from currenty-available hardware, which people interested in working with 3270s would find of interest. Peter Flass (talk) 03:45, 16 May 2021 (UTC)

Provenance of downstream load (DSL)
added the text The 3179G was the first terminal that required to download its microcode from the cluster controllers floppy or harddrive.. However, the 3290 was announced in 1983 while the 3179G was not announced until 1985. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 15:59, 8 June 2021 (UTC)

Announcement letters
I noticed that the 3179G sections cites the announcement for the 3179 rather than that for the 3179G and that the section for the 3290 gives the wrong title for the announcement letter. I'm in the process of compiling from https://ibm.com/common/ssi a table of IBM hardware that will include links to the relevant announcement letters; that will include all of the 3270 announcement letters that I can find. Once I've added all of the 3270 dates that I can find, I plan to include them in the article with links to the announcement letters. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 20:10, 9 June 2021 (UTC)

Organization of 3174 controller section
A recent edit added IBM 3270 after IBM 3270, severing the historical information on the TN3270 support from its proper context. I propose that #3174 controller be restructured as a description followed by separate subsections for configuration support and history. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 14:24, 16 June 2021 (UTC)
 * Sorry, That was definitely not the intention, I will fix that soon.
 * I forgot that IBM used the term configuration support for the 3274 as well; I can't find the announcement letter for the 3274-1A, 1B and 1D, but you can see the term in the announcement letter for some later models. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 22:43, 17 June 2021 (UTC)
 * I forgot that IBM used the term configuration support for the 3274 as well; I can't find the announcement letter for the 3274-1A, 1B and 1D, but you can see the term in the announcement letter for some later models. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 22:43, 17 June 2021 (UTC)
 * I forgot that IBM used the term configuration support for the 3274 as well; I can't find the announcement letter for the 3274-1A, 1B and 1D, but you can see the term in the announcement letter for some later models. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 22:43, 17 June 2021 (UTC)

3274
@Chatul: A recent edit of 3274 states that “The 3274-1A was an SNA physical Unit type 4 (PU4),” Are your Shure?, in my book it is a PU Type 2.0, the FEP is Type 4. - unsigned comment by AndFrog.


 * I could be mistaken; if you have a source, please change it and include a citation to your source, including page, section, url and section-url as applicable. If it's a PDF then use &lbrack;url#page=PDF-page page&rbrack;


 * Have you thought about the best way to document configuration support for both 3174 and 3274?


 * Please sign you comments with -- --Shmuel (Seymour J.) Metz Username:Chatul (talk) 16:09, 21 June 2021 (UTC)


 * On second though, you're definitely correct because IBM later added APPN support, which implies PU2.1; I don't recall whether they also added it for the 3274. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:29, 21 June 2021 (UTC)

Code charts
The character code charts appear to be for the 3277, and do not include the buffer orders for Extended Data Stream, e.g., SFE. --Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:08, 14 November 2021 (UTC)