One of my biggest headaches as a DM (though perhaps I am overcomplicating things) is coming up with consistent prices for goods and services not listed in the reference books for a medieval setting (or any setting for that matter) campaign I'm running. Does anyone have any rules of thumb, or better yet, some overarching online encyclopedia (...in my dreams :-P) for the cost of items/services and how they vary them for settlements/villages/towns/cities in times of peace/war/etc?
Generate them randomly. You'll save yourself a lot of grief, and you won't have to spend a lot of time elbow-deep in rulebooks. Assuming a typical faux-medieval system of copper/silver/gold...
Could any peasant get this? Then its price is in copper. Would the resources of a town/city be needed for this thing to happen? Its price is silver. Is it something you'd expect to change the course of an adventure, like a weapon or a spell? Its price is in gold. Highest value coinage takes precedence.
Roll 2d20 times the level of the relevant PC. Bingo -- you have a price. It may not be consistent from a world-economy-simulation standpoint, but it will give players a reasonable expectation that will grow with their characters' experience.
Some of the suggestions here hypothesise a complete capitalist economy in goods, with goods being made speculatively for transparent consumer markets.
Medieval economies don't work that way. A "capitalist" economy might make sense in some conditions: medieval Islam, early modern Holland, a backwards projected "science" fiction such as Q or The Baroque Cycle, or pure fantasies about modernity but with big swords. A "capitalist" economy doesn't make sense if the basic form of production is handcraft domestic production by land bound agricultural labourers who are owned, or semi-owned by being land bound to owned land, by a nobility. Feudalism and capitalism don't mix as the fundamental forms of human production. When they do mix, it is in strange and aggressive forms (see the British invasions of India).
So when trying to price silver broadswords, or wands of dancing, or mime services, I suggest two things: *Big ticket items cannot be purchased at all. They must be granted as boons as the result of services to persons of high station (in what ever fundamental sequence of magic user, priest, fighter or rogue is applicable). *Small ticket items need to be purchased over the course of weeks or months by seeking out fairs or by "direct ordering" from makers of goods.
In both cases, the gold costs involved to locate such items should be used to drain off the "hoard" effect of uncovering mysterious and lost treasures. Pricing should be story driven to the extent that the cost of an item is to make a character as poor as you need them to be to make the player have as few options to resist doing the next thing you want them to do.
In simulationist terms, this is what happens to hoards in medieval economies. See what happens, for example, to soldiers loot which dissipates like fairies with the morning.
If you're still fixated on doing a full input-output table of multiple commodities and productions (and it suits your economy anyway, your base setting is Islam at 1000, or the Dutch Revolt), then I strongly suggest not doing it. This is an incredibly boring approach to computing commodity prices, and while it is very possible to do so (both the Soviets did it by hand with a couple of hundred "commodity" representations, and it is possible to do it with thousands, hundreds of thousands, or millions of commodity bundles on home computing) the pay off is boring.
If you take labour-input prices (given that you need a wage economy in the first place, so that's Islam out, and the Dutch too really..., so lets set it in Randian Planescape), then your key problem is actually uncovering the labour time and production time of goods, which is boring. If we followe EP Thompson on the development of capitalist time ( Time, work discipline, Industrial capitalism http://past.oxfordjournals.org/content/38/1/56.citation ) then we notice that workers resisted having the labour time for production reduced to the minimal labour time required for production in any one era. When you order that silver broadsword, even if it takes 5 weeks under the prevailing system of production at best, the smith will lazily take 15 weeks.
So: due to friction effects of wages not existing, markets not existing, goods being made to order as a result of reciprocal fealty relationships, actual labourers resisting work discipline and time intensification and instead enjoying work, and the dramatist need to restrict player action to give them something to hope for: set prices arbitrarily high, require adventures for special goods (including "normal" items that someone else would need to make to order), and set lengths of time arbitrarily high. Unless you're keeping track of some social status indicator, in which case the "hero of X" or the "saviour of Y" or the "queen of Z" might get a sword made special fast for her.
Go beyond the PHB. There are a lot of supplements out there that have been included in other lists, such as those on the d20pfsrd.com site. Unfortunately, D&D does not current have, nor has it had, any kind of realistic economy. It doesn't really matter what scale you're looking at either. That pretty much leaves you with two options:
A more realistic view on tradesmen and craftsmen
You need to start with a fairly complete Goods and Services list, otherwise you end up having to assign raw prices yourself.
From there, you can extrapolate on cost based on raw minerals and materials (see the trade goods section of the link above). Once you have an idea of the raw cost, you can then start to consider what individual craftsmen and merchants might want for their time and crafting abilities.
This all assumes that you want as realistic of an economy as you can possibly have. I don't believe that D&D has a very realistic economy. It's more an "accidental economy" based on the requirements of the game being balanced.
Off the cuff
As a GM, this is the way I normally go, because I don't generally have all the time in the world to consider every option. Most things that a player will want can be compared against several existing objects. What you do is "triangulate" an amount of money based on as many existing products as you possibly can. That way you have a balanced end product against what the worlds current economy looks like.
Going with the more off the cuff method does tend to result in a bit better balance for the game world, because that way you're not trying to apply concepts of economy where they don't generally exist.
Look at the various Ars Magica suppliments, especially City & Guild
While this answer does not address 3.5 specifically, the level of research put into Ars Magica supplements and their generality makes them an excellent resource for your specific problem. City & Guild covers costs, manufacturing, and the nature of guilds in cities. While there are a few pages of crunchy-for-Ars stuff, most of it can serve as an exemplary encyclopedia.
If you can deal with the grognardism at the Tao of D&D, there are, buried in the archives, ways and means of simulating entire world-economies.
At the end of the day however, the 3.5 economic system crumbles if you look at it too closely, so I wouldn't worry about prices and charging. See: price of ladders and ten foot poles.
May I suggest the supplement "...and a ten foot pole", originally for Rolemaster but probably one of the best guides to what things should cost in various eras.
Hodges List of Prices is a shorter but free resource that can help you form a more complete idea of pricing.