This discussion comes up frequently at LatinTeach; that was what prompted me to start trying to collect some links to online texts with macrons. I've pasted in below my thoughts about that; you can join the LatinTeach listserv here if you are interested: http://nxport.com/mailman/listinfo/latinteach
-it's a pretty active list. The people who consider macrons essential are the people devoted to the re-creation of Roman pronunciation - the phonemic distinction between long and short vowels was already gone in proto-Romance (except for the lingering results in word stress patterns), and ceased to be phonemic in the Romance languages, so the advocacy of macrons is a very modern phenomenon, linked to the 'restored' pronunciation movement. For notes against the restored pronunciation movement, see Bennett's retraction of his earlier advocacy: http://www.memoriapress.com/articles/Bennett.html
- although I should add that my argument against macrons is not an argument against pronunciation (I don't care what style of pronunciation people use: CUIQUE SUUM); my argument against macrons is purely about Latin orthography and how best to promote confidence in reading unmarked Latin texts.
I've seen that for many students macrons provide just the opposite of the effect people here are hoping for: instead of being a spur to learning, they instead become a crutch. Teaching college with students coming straight from high school where everything was marked with macrons, I saw many students who were painfully unable to resolve ambiguous forms on their own, and instead assumed that things without macrons were short. Well, that's a recipe for disaster.
Given that you will find relatively few Latin texts marked with macrons in the larger Latin world, people need to be prepared to read texts without macrons confidently. So, I'm all for marking first year texts and reference books like grammars and dictionaries, but I am far less persuaded that marking all texts that students see is a good idea. Macrons are not part of the written Latin language; they are a modern invention. If the Romans had wanted to distinguish in writing between short and long vowels, they could have done so (writing the vowel twice, as some languages do, or using separate characters for the long and short vowels as the Greeks did for some of their vowels). Instead, the Romans opted for indicating vowel length only in extremely ambiguous situations; it is a very rare practice in Roman writing. This was the case for written Latin up until the 19th century. Punctuation has traditionally been added to Latin texts (although often at variance with the punctuation we expect in modern English, reflecting instead different times and national traditions), but macrons are something rarely found in printed Latin texts outside of the very narrow world of modern school texts.
Compare the language teaching practices in other languages. Russian, for example, is notorious for its variable word stress. As in English, Russian stress is quite unpredictable and one of the great tasks that faces any student of Russian, like any student of English, is coping with the importance of word stress in the phonology of the language, but the absence of any written markers indicating the stress. In first-year Russian textbooks, stress is marked (but even then only some of the time) and it may be marked in basic readers (but not always)… but beyond that, it is not marked, because students learning Russian really need to learn Russian, as printed with the Russian alphabet, which does not include stress marks. So too with English - stress is not marked in writing, even though it is a core feature of the spoken language.
I won't raise here my own personal lack of interest in teaching students to use diphthongs in a pseudo-imitation of Latin phonemic vowel length (a separate topic) or my own preference for marking stress rather than macrons (a style found usually only in ecclesiastical texts, but which is suitable for any printed Latin text and which is a much bigger confidence booster than macrons I have found in prompting students to read out loud), but I do want to note that if you want students to be able to take advantage of the range of printed Latin texts over the past centuries, they need to learn to read without macrons, resolving vowel length ambiguities on their own, rather than expecting the editor of the text to do that for them by adding all the macrons.