Talk:Green Hills Software

From Wikipedia, the free encyclopedia

Articles for deletion This article was nominated for deletion on 10 June 2007. The result of the discussion was keep.

I don't know why this reads like an advertisement, so I removed the tag from the top.

Contents

[edit] Stubified

I have reduced this article to a stub. The so called 'Products' was a total unsourced mess. This can now be expanded with /sourced/ material. BlueValour 21:02, 27 August 2006 (UTC)

The unsourced material has been added back without sources. Please only add content back if it is accompanied by a source - see WP:CITE. BlueValour 20:52, 23 September 2006 (UTC)

See also from WP:V "The burden of evidence lies with the editors who have made an edit or wish an edit to remain. Editors should therefore provide references. If an article topic has no reputable, reliable, third-party sources, Wikipedia should not have an article on that topic. Any edit lacking a source may be removed ..."

The unsourced material is: '*Optimizing compilers targeting a variety of 32-bit and 64-bit platforms, including ARM, Blackfin, ColdFire, MIPS, PowerPC, x86, and XScale.

  • Hardware debug probes for platforms including the above.
  • MULTI, a multiplatform IDE for C and C++ able to run on Windows, Linux, and Solaris. Aimed at embedded engineers, it is tightly coupled with Green Hills' optimizing compilers and hardware debug probes, and therefore can be used to develop for all the processors in the above list. Additional features include an integrated CVS browser, a diff viewer, automatic code completion, graphical class hierarchy generators, integration with Eclipse, a bug tracker, and scriptable breakpoints.
  • TimeMachine, a set of tools for optimizing and debugging C and C++ software. TimeMachine records every instruction executed on a CPU, archives the instructions, and allows the developer to review the executed instructions. (In other words, the debugger can trace or single-step "backwards in time" as well as forwards; hence the name of the product.) On embedded processors, TimeMachine is implemented using a trace port on the CPU. Since trace ports are built directly on the processor die and thus have virtually no performance penalties, TimeMachine can collect debug information at full speed.

    TimeMachine is useful for analyzing race conditions and other Heisenbugs. The ability to replay instruction sequences at a later time is also useful for embedded engineers who cannot use breakpoints because halting the program is impossible (for example, when debugging the flight controller on an airplane).'

BlueValour 21:01, 23 September 2006 (UTC)

The external link points to GHS' Web site, which contains extensive marketing materials. In my revision of the Wikipedia material, I tried to remove some weasel words and subjectivity, to make the article encyclopedic in tone.
I will restore the material you deleted tomorrow, unless you can convince me that you really don't believe Green Hills' Web site, their marketing materials, or the common knowledge of their employees and customers. There aren't any trade secrets in the Wikipedia material that I can see, so I don't think there are legal issues. Please elaborate: why do you oppose this material? --Quuxplusone 00:00, 24 September 2006 (UTC)
The provision of an external link is not sourcing. It is not my job, nor that of any other reader, to try to validate statements in the article; it is your job to source them, linking each statement with a website reference. "the common knowledge of their employees and customers" is OR and not admissable. If you reinstate material without sourcing it will simply be liable to be removed. A more constructive approach would be to recast the material into a sourced form or suggest an alternative. BlueValour 02:26, 24 September 2006 (UTC)
I really, seriously do not understand how the "unsourced form" of this article is any different from the "sourced form" of the article on Angela Lansbury. I've added links to source material anyway, but I think they make the article look crass and cluttered. Wikipedia is not an extension of a commercial site. If you think Green Hills isn't notable or encyclopedic, just nominate the article for deletion; I'm a sometime deletionist myself, and I will abstain from voting in this case, because I think its notability is borderline. However, please stop reverting this article to a meaningless stub. If you think there's something wrong with it other than the fact that it's about a commercial company, fix it. If the commercial nature is the problem, nominate it for deletion; Template:prod is thataway. <--- Thanks. --Quuxplusone 05:31, 24 September 2006 (UTC)
BTW, you might take a look at other pages in the same category: Wind River Systems, LynuxWorks, MontaVista. Maybe you could clarify why you aren't reverting them to stubs too. (I'm not trying to make an argument that "everyone else is doing it!", just asking why you believe Wikipedia standards require more than is already present in the current article.) --Quuxplusone 05:55, 24 September 2006 (UTC)

[edit] Moving towards NPOV ...

Quuxplusone has reverted some of my edits to the GHS text. Many of Quuxplusone's edits are very good. Others I do not agree with. I am an journalist covering embedded tools for a professional electronics magazine. GHS is one of the companies that I talk to and write about, since it is has a big customer base and is often innovative in its products.

Some points where Quuxplusone changed me, where I don't agree:

  • The term "Optimizing Compiler" is, I think, just a marketing term. All compilers optimize their code.
  • The list of competitors seems generic and imprecise. It is very similar to the other embedded tools vendors competitors list. It also only adresses competitors for the RTOS products and not the compilers and probes, which seems assymetric. I actually do think it would be nice to have a paragraph on competitors, as there are interesting things to be said about this -- there are not many 178B RTOS:es to choose from, for example.
  • "royalty free". This is a slogan GHS uses to describe the pricing of its products. But I do not think for example that these products are license free. So, if you introduce info on pricing, like "royalty free", please be complete in that info and specify for example that there is a license cost, if there is one, which I think there is.
  • "low cost". Compared to what? Other USB probes? I think "low cost" quite obviously should be deleted. The alternative would be to add "high cost" to describe the other probes ... ;-)
  • "ANSI C, including all the features of ANSI C" is a tautology.
  • "multi platform" -- almost all the IDE:s in this business are, so this is in practise redundant. In fact, the interesting platform issue about GHS Multi is that is not Eclipse adapted. I think just about only GHS and Microsoft are not Eclipse now.
  • Awards. Well. Hm. They all have some award to show up, or another, I think. So this does not say much.

My basic point: in order to be included in wikipedia it is not enough that a fact is simly true. It has also to be relevant.

For example, if wikipedia says that INTEGRITY is "POSIX certified" as opposed to just "POSIX", then wikipedia claims that the certification is relevant. I do not think it is. At least not relevant enough to be included in the article. Not until, for example, there is an article in the wikipedia describing the certfication process in point.

--gnirre 17:16, 10 July 2007 (UTC)

My comments, point by point:
  • "All compilers optimize their code" — Mostly true these days, but not in the sense I was trying to convey here. Green Hills' major claim to fame (so to speak) is that its compilers really optimize code (as in, make it run faster and occupy less ROM) better than anyone else's. Even though I agree with you that almost every commercial compiler these days technically includes optimizations, you wouldn't expect to see Microsoft Visual Studio, for example, advertised as an optimizing compiler! Optimization should be a key point in the section on compilers; I don't feel qualified to write a paragraph on it, but I think the word "optimization" should at least be mentioned, for now. :) Maybe EEMBC benchmark scores should be mentioned somehow, although I think that would be drifting too far into tech-biz-cruft. [1]
  • Royalty-free: Defined by GHS as "No run-time royalties (payments) per deployed copy of the RTOS."[2] If this phrase is indeed unique to GHS, I think that should be noted in the article. I really have no idea how common the idea is, or what other software vendors call it. (I think it's slightly more than bizspeak for "You can get a site license", because copies of an embedded RTOS could end up literally all around the world, whereas site licenses are normally geographically limited.)
  • "POSIX-certified", however, appears to have a well-defined meaning.[3]
  • "Low cost" compared to the other probes, I assume. If we're going to mention that GHS sells N different kinds of probes, we need to mention how they're different from one another. It makes more sense to say "Probe A is a fast probe and Probe B is a cheap probe" than to say "Probe A is a probe and Probe B is also a probe."
  • "Features" of ANSI C in the sense of "benefits" or "extra permissiveness compared to K&R C". If you have a better word, please share; I've actually debated the meaning of the word "feature" on Wikipedia before, taking the side you're taking now. :) The point is that "permissive mode" is not identical to ANSI C, because it relaxes some restrictions; but it doesn't remove any of the positive (non-restricting?) features. Actually, I think the whole "five dialects" angle could be discarded; the real point is that GHS supports MISRA C, K&R C, ANSI C, and mostly–GNU C (as well as C++ and Ada), and "five dialects" is kind of arbitrary.
  • I'm willing to let "multi-platform" and "Awards" disappear. I think the award is relevant, but I see several other tech-company articles lack that kind of information, so I guess it's only fair. --Quuxplusone 06:30, 11 July 2007 (UTC)


The homepage of GHS serves a completely different function from the Wikipedia. The purpose of the GHS homepages is to sell GHS products. And this will always manifest itself in the facts that GHS puts forth on their homepage. And also, equally important, which facts they do not put forth. You cannot possibly get a good wikipedia text on GHS in the Wikipedia by only using those GHS homepages as your source. Same points by point:

  • If you can verify through sources that the following quote of yours is correct, then my suggestion is that the headline would still be "Compilers", but that you should add exactly that quote to the text: "Green Hills' major claim to fame is that its compilers really optimize code (as in, make it run faster and occupy less ROM) better than anyone else's". I can offer you another way out of this: get the facts on the EEMBC scores, and put them in the text.
  • What is the point of telling your customers that you will never, ever sell to them by royalty? Do you believe royalty is actually always a bad thing for the customer? It is not. Royaly could for example be used as a tool for risk sharing. So actually, I would be ready to describe "never roylaties" as a limitation of the GHS offerings. Unless you can explain to the readers of Wikipedia what the point and relevancy of "no roylaties" is, then that fact should not be in here. You should at least be able to explain and defend if "no royalties" is a limitation or if it is a feature.
  • Reading your link I find that "Posix certfied" does NOT have a well defined meaning. "1003.1TM-2003 Base certification" on the other hand, has a well defined meaning, together with two other certifications. But still, my objection is: is this *relevant*? I want you, Quuxplusone, to argue for why certfication is relevant? As with "roylaty free" I don't think this is obvious.
  • "Low cost" has no meaning at all. "Lower price that GHS's other probes", that means someting. But again, is it relevant? I am trying to think of a context where this information would relevant. One case would be for example, if GHS was trying to sell their products to hobbyist users. But they are not. They address producers that are in the process of constructing stuff to sell on markets. Give GHS a phone call, and they will probably give you any of their probes for free, if it could mean that they could sell you a license for one of their OS:es. When GHS write "low cost" on their pages, they are only using market speech. The phrase should obviously not be copied to Wikipedia.
  • I still say it is a tautology (this is not so important to me, though)
  • Awards, multi platform ... (check, check)

--gnirre 15:04, 13 July 2007 (UTC)

[edit] Continued

Clearly we're getting close to an edit war here. :( Stop removing the "Competitors" section; it's necessary and matches the similar sections on all other articles in this series (Wind River Systems, MontaVista, et cetera). I may be touting GHS' probes as "low-cost", ;) but at least I'm not removing wikilinks to alternative suppliers! You seem to be removing the keywords "royalty-free" and "-certified" on sight now, without even bothering to check the surrounding text for grammar or related extlinks; I've put the words back. You've also deleted an en dash (long hyphen) between the words "ARINC-653-1" and "compliant" several times without explanation; the word "ARINC-653-1–compliant" is a compound adjective, and requires the dash.
GHS' compilers do not target all 32- and 64-bit platforms, and it's misleading to claim so. Hence my phrasing "a variety of". Likewise, AFAIK they don't support all GNU extensions (although I admit I don't know any missing ones off the top of my head); hence the qualifier "many" (i.e., more than "some", but not "all").
MULTI is aimed at embedded engineers. ("Engineers working in the field of embedded systems" might be a less confusing phrase?) This is as distinguished from many comparable IDEs, which may be aimed at desktop application developers or Web developers. That's not saying MULTI is better; it's just identifying the product's target audience as context for what follows. The CVS browser is integrated with the debugger and editor; it's actually quite cool, if I might say so myself. (I.e., the system does not include "a CVS browser", but rather "an integrated CVS browser". Many instances of "a CVS browser" are available free; GHS probably couldn't get away with selling just "a CVS browser".) --Quuxplusone 01:02, 14 July 2007 (UTC)

The basic problem with this article is that it is written like an advertisement for GHS instead of being written like an encyclopedia article. For some reason, you don't seem to see this. The article was even recently suggested for deletion for this reason.

I have two questions for you: 1) Do you Quuxplusone represent yourself, or are you acting on someone elses behalf, for example GHS? 2) What is your comment to my statement that you cannot get a good article on a company for Wikipedia just by copying facts from a company's homepage?

  • Regarding the phrase "intended for use in embedded systems requiring reliability and fault tolerance" that I am removing, and you are putting back: could you find me an example in the Wikipedia of an operating system described as "intended for use in embedded systems NOT requiring reliability and fault tolerance." My point is that until you give the reader a clear statement on just how INTEGRITY is "reliable" and "fault tolerant", it gives the reader no information to have that phrase there. A description of the nature of its fault tolerance and reliability, would also actually probably make very interesting reading.
  • Your Competitions text is not good. You mention RTOS products, and not probes, IDE:s or compilers -- why this assymmetry? You do not separate between INTEGRITY and Velosity, which are quite different markets. The list also mentions CE and Linux apart from the others cometing OS:es, without explaning why, and also describing them derogatively as "lesser" competitors. A text on the competition on this market could be very interesting, but that text as not been written.
  • You put back "roylaty free" without commenting on my discussion on why I removed it. This is not fair.
  • You put back "certified" after POSIX in the same manner.
  • Hyphenation -- if you say so, sure.


(I added a "Continued" subsection above so there'd be less to scroll through in the edit window.) (1) Of course I "represent myself", and not GHS or Elektroniktidningen or Wikimedia or anyone else. (2) Certainly it's unlikely that a good article can be gotten just by copying facts from a company's homepage, and while the GHS article isn't "just copied", I agree that it does rely too heavily on promotional materials. (However, the solution isn't to remove the few tidbits, such as "Competitors" and "Awards", that do have reliable sources. That just makes the article even less reliable.)
"Why this asymmetry?" Why not? I didn't write that text in the first place; I just object to its wholesale deletion. Anyway, I don't think GHS has any across-the-board competitors in the compiler field; its competitors there tend to be the target-specific compilers produced by the chip manufacturers (Intel, ARM,...) Other IDEs would be Microsoft Visual Studio and random other editors; editors IMLE tend to have less "markets" than "individual fanatical devotees." ;) I don't know about probes; do you? "Lesser competitors" in this context obviously means "entities whose products or services are less directly competitive with Green Hills'", not "entities who are morally inferior to Green Hills"; do you think this would be unclear to the average reader?
We've discussed "royalty-free" and "POSIX-certified" already. You didn't notice, but I did add the links from this Talk page to the article text — one to GHS' definition of "royalty-free" and one to the POSIX organization's Web site certifying INTEGRITY and giving specifics of its compliance with the POSIX definition. You can't argue that these terms are poorly defined or irrelevant; they are well-defined (see links), provide context, and differentiate the OSes from one another. (See above on "Probe A is a probe and Probe B is also a probe.")
Think of it this way, please: Every edit you or I make should be making Wikipedia better. Leave articles better than you found them — whether it means fixing grammar, or the flow of a sentence, or the paragraph organization of the article, or adding missing information, or removing trivia. Be alert and make sure that doing one of these things doesn't harm any of the others. If you can't fix a sentence without changing its meaning for the worse, it's better to leave it alone. If you can't remove trivia without confusing the sense of the article, it's better to leave it alone. I don't object to rewriting a problem section or article from scratch... so long as the new version is better than the old one! (The "Show History" corollary is: Leave articles better than they have ever been before.) --Quuxplusone 21:20, 14 July 2007 (UTC)

[edit] Fixed Competitors a bit

Removed "having a hard time fighting linux". It was unsourced and it isn't clear that Green Hills is having a hard time at all. Removed an unsourced claim that linux is the de facto standard for embedded. On the contrary its still rare compared to no OS. Removed an unsourced claim that people use linux for its GUI and TCP/IP features. That isn't true. I added more own unsourced claim that they use it for its advanced networking features (routing, NAT, etc). —Preceding unsigned comment added by 63.102.70.70 (talk) 03:17, August 26, 2007 (UTC)