Macrons are not part of "standard" Greek orthography (meaning you won't find them in a typical printing of Plato). They are used, however, in Greek studies as a learning aid, to mark non-obvious long vowels. The only vowels that ever require them are alpha, iota, and upsilon, and then only when it is not clear from the accentuation of the word whether the given vowel is long or short. If one employs this system with consistency, then there is no need for a separate breve mark for short vowels because one knows that any alpha, iota, or upsilon that doesn't have a macron (or a circumflex, or that is not clearly otherwise long) has to be short.
There is some debate about whether macrons should be used at all, and if so, when and where. Most agree that they do have (at least) limited pedagogical usefulness, especially for studying Greek poetry, and for beginners learning morphology for the first time. They are also (most would agree) a necessary component of a good lexicon, since that is where the definitive forms of the words are presented. The debate essentially focuses on how frequently, and for how long, a teacher (or grammar book) should continue using macrons in the presentations of textual examples (exercises, readings, etc.).
Macrons are most useful when learning morphology. Certain inflectional endings have either long or short vowels, which affect accentuation. If students learn these from the beginning, they are more likely to remember them throughout their later studies. In addition to this basic use, macrons are also generally considered indispensable when studying Greek verse, especially its composition. One must know the quantity of vowels in order to understand, analyze, and generally appreciate (intellectually and emotionally) a Greek poem. As for general use in reading passages (which are designed to prepare students for reading general Greek texts), the use of macrons can be justified using the "training-wheels" analogy. Just as most reading passages in grammars provide glosses for unfamiliar vocabulary (which has yet to be introduced), so also does it seem reasonable to provide vowel quantification "glosses" as a further aid. One hopes that the students understand that such macrons will not be found in "live" texts.
A place where using macrons is imperative (as was recently pointed out to me) is when working in a group (like on the Internet) that is studying poetry because communicating one's intended (in the case of composition) or interpreted (in the case of reading) vowel lengths is one of the main aspects of that type of study. Such a group environment really does require this type of communication, and macrons work for this purpose, in a consise and acceptable manner. Another very good point is that the consistent use of macrons in one's own study can help to reinforce the vowel lengths in both common and uncommon words, which will then, when reading "live" (macron-less) texts, make it easier to recognize (remember) the vowel lengths where necessary.
The primary argument against macrons is simply that they are not present in real Greek texts, so students will eventually need to get used to reading without them. Ambiguity is present in language at all levels, and students need to learn how to deal with resolving these real ambiguities. Grammars should gradually wean their readers off of the macron milk, and offer methodologies for learning to recognize and resolve ambiguous situations that result from uncertain vowel length. This is not presently done in any of the many Greek grammars that I have read. Of course, "real" Greek texts don't have lower-case letters, spaces between words, line breaks falling cleanly between words, or even half-way decent spelling, and they generally do contain very poor handwriting. So there is something to be said for these modern editorial niceties.
If you're not a techno-type person, feel free to ignore this point, as it only has bearing on the use and presentation of Polytonic Greek on computers. We all know that Greek on computers is beyond annoying. The Unicode consortium has helped a great deal with its encoding of the precombined Polytonic Greek characters. This process, however, has had problems from the beginning, when they initially coded iota subscripts to be placed below capital letters. Presently, all the precombined characters used in "standard" Polytonic Greek texts are sufficiently represented. What remains to be done is to include some of the non-standard (but frequently used) combinations. I think the first new combinations that need encoded are those including macrons.
Currently, there are only the three basic combination characters with macrons (over the three vowels). If one wants to add a macron to a vowel that already has other diacritical marks, the only way to do it is with a combining diacrital character. This technically could work, if there existed fonts that could handle this adequately in the most common cases. However, none of the currently available fonts are capable of displaying all the possible macron combinations with any degree of acceptability. The macron is either too far to the left (in the case of iota), or else it obscures the existing diacritical marks. These facts are too annoying for me personally to use macrons in my own presentations. For others, it does not bother them as much, and they are content to use the given technology, believing the benefits of macron usage outweigh these annoyances.
At the moment, I don't have any good, comprehensive solution. I'm working on editing a font to try to make the combining macron work better. In cases where the vowel has no other diacritical marks, there is no problem using the precombined macron character. In other cases, I either put the vowel-with-macron precombined character in parentheses or brackets following the word, or else I bite the bullet and use the combining macron, despite its ugliness.
Overall in this debate, I tend to lean toward the "con" side, preferring to avoid macrons when at all possible. This is because I do not want to come to rely on them as a crutch, and also because they are (technically) annoying. I also have not ventured very far into the poetry world, so have not felt the need for them that I otherwise might. As I progress farther into my own Greek studies, I plan to read more poetry, at which time I may need to reassess my own position. With luck, by that time, the technical annoyances will have been solved.