I don't think it is very plausible that the holodeck may create fully functional weapons in the first place, and only a safety protocol prevents them from being dangerous or lethal. While we know that there are other fully operational holographic devices too (like the generator in TNG: "A Matter of Perspective" and the interface Barclay built in TNG: "The Nth Degree"), I wouldn't think that the same applies to possibly lethal weapons. I would much rather expect that the holodeck only simulates the effect a weapon or any other action has on a holographic target, but not on anyone or anything physical that is taken inside the holodeck. For a battle simulation it would be useful to see these effects without doing any real damage, but no one in Starfleet would ever seriously consider to severely hurt or kill someone in a holodeck. Least of all, I would expect the holodeck to simulate a fictional weapon like the "Captain Proton ray gun" which is supposed to be only a prop in Star Trek, just like Star Trek phasers are in the real world. The same should apply to Picard's machine gun, or did he ever want to use them against real targets?

If we assume that the holographic weapons are fully functional according to some set of parameters though, it would not matter if they were perfect simulations or the real thing - a replicated weapon that would work outside the holodeck too. Replicated weapons would face only one serious problem: The holodeck is a lot smaller than the distance between the weapon and the target may appear to be, and maybe it is not even physically directed towards the target. There would have to be some sort of forcefield or transporter involved to carry the bullet or whatever is fired to the location of a real target within the simulation, just like the "treadmill effect" for persons inside the holodeck.

Shields & Cloaking

Shapes of shields Since TNG we are accustomed to the typical "shield bubbles" around a ship. It doesn't seem to be realistic in the first place to have a sharp border between the shields' inside and outside, if shields have anything in common with real forcefields, whose potential always decays with the reciprocal first power (r^-1), and its field strength with the reciprocal second power (r^-2) of its radius. This applies to electric fields, as well as to gravitational fields, the latter of which are supposed to be the basis of a Federation ship's shields, according to the TNGTM (see also "shields and cloaking" further down). The typical shield bubble since TNG was an ellipsoid, as it would be created by shield emitters sufficiently far away. Such a bubble can be seen very clearly in "First Contact", when the Borg lock on the Enterprise-E.  The shape however has changed recently, as we have seen the Enterprise-E with separate shields in "Insurrection" (that lit up only in a particular region when hit), and finally with almost "skin tight" shields in "Nemesis". It doesn't seem like an improvement (well, it looks rather like an improvement of visual effects), if incoming fire is deflected or dispersed as late as it has almost reached the hull. Also, the tight or partial shield would not have the homogeneity of the bubble, therefore exhibiting weak spots, even if crucial areas may be better protected. It is questionable whether the Federation would change its technology so considerably.

Shield percentages What does "shields at 47%" mean? Is the average shield strength down to 47%, or is it rather the angular coverage? It seems that shields may not be switched on and off instantaneously, they may need to be built up and shut down as seen in "Star Trek II". They may be charged in a similar way as phasers. In this light they are not devices that always operate at 100% (or even more), if only they are provided with enough power. Rather than that, they appear to discharge and thereby to lose their strength upon impact, even if nothing on the ship has taken physical damage. This would corroborate the notion that the whole system of shields, which forms a continuous bubble (see above) loses a certain percentage of its overall strength, no matter where it is hit.

If, on the other hand, the shield percentage were to refer to a coverage, this would imply that the shield emitters in certain areas failed completely, which would make no sense if there were no considerable other physical damage. One of the few arguments in favor of shield coverage is that power may be diverted to the shields. This suggests that, with a part of all shield emitters gone, it may be possible to increase the protection again by powering up the rest to more than 100% (there is always a safety margin in engineering). But this doesn't seem to be the case, as we have always witnessed shields continuously losing their strength. It seems that the additional power would be spent rather to recharge the shields than to gain immediate strength through more power. If a percentage of 50%, for instance, stood for a coverage, it would mean that half of the hull surface were totally unprotected. In other words, the ship would make quite a target and would take disastrous damage with the very next hit, which usually doesn't happen either. Instead of that, the percentage keeps further decreasing, while the ship remains intact (with the exceptions discussed in "damage despite shields" below).

There seem to be better arguments that shields lose overall strength. But the recently frequent mentions of particular areas like "aft shields down to..." seem to contradict this theory. In this light, maybe the shield strength should indeed take into account weakened areas as well as its overall strength. It is possible that the percentages are a combination of both the shield strength and coverage, calculated with a rather complicated algorithm. Only in case that there are particularly weakened areas that the captain would need to protect, the tactical officer would refer to something like "aft shields" in addition to the total figure. 

Damage despite shields Ships in Star Trek seem to take considerable damage already when the shields are still up. First of all, the ship rocks every time it is hit, although the shields can't be supposed to form a rigid framework around the ship that would be mechanically coupled to the hull. On the other hand, since the shield around the ship is changing its geometry (or is "pushed aside") when hit by weapon fire while the shield emitter power remains the same, this could indeed result in a force on the hull. Even simpler, the shaking could arise from a smaller amount of weapon power that is able to pass through and reach the hull.

The usual effects of weapons that penetrate the shields are scorches on the hull (most obviously in "Star Trek VI"), but very often even exploding plasma conduits inside the ship (as about every time when the Defiant or Voyager is attacked). It is yet another issue why high-power conduits are unwisely routed through bridge consoles, as there may be all sorts of security concerns, not only in a battle. Moreover, the most exposed location of the whole ship doesn't seem to be the right place for the bridge. Anyway, the question arises why and how exactly this specific kind of damage frequently occurs. Still assuming that "holding" shields keep away most of the weapon's energy, we should assume that, if at all, only the hull would be affected, and that the metal should shield the interior like Faraday's cage. Therefore it is not evident why most of the damage seems to occur inside the ship. Maybe the metallic hull plates actually have no effect in blocking weapon energy which we may explain with some quantum physics and technobabble. At least, it wouldn't matter in such a case where the bridge was located.

Simultaneous shields and cloaking There is no generally valid answer to the question if a ship can be cloaked and shielded simultaneously. Romulan and Klingon cloaking devices seem to work similarly, although the ones of the Romulans are generally considered to be more advanced - also because, at the time of TNG, the Federation knows a bit about Klingon technology, but almost nothing about that of the Romulans.  In TNG: "Redemption" we saw cloaked Romulan ships being fired on, and the weapons fire was dispersed, which indicates that they did have the shields up. In TNG: "Face of the Enemy", however, the Romulan Warbird couldn't be cloaked and shielded at the same time. The same applies to the Defiant in DS9: "The Search". The Klingon D-12 Bird-of-Prey in "Star Trek: Generations" could not keep the shields and cloaking up simultaneously either, and Worf used exactly this weakness to get the shields down and the cloaking up to be able to destroy the ship. At least the Klingon ship may easily be explained in that it was explicitly called outdated anyway, but the question about the capabilities of Romulan ships remains.

There is a more general problem about cloaking and shields if we take into account their working principles. The shields of Federation starships, according to the TNGTM, are made up of a graviton field and not, as one might have expected, of an electromagnetic field. This way, the shields would be based on the same technology as the cloaking that employs a graviton field to "bend" light rays around the ship, and both fields would possibly obstruct or even cancel out each other. On the other hand, we don't have to believe everything in the TNGTM if not explicitly stated in screen, and even if it is correct for Federation starships, the Romulans or Klingons may easily use a different shield technology that doesn't interfere with their cloaking. Maybe the similarity between shields and cloaking is even the explanation for what we have seen in "Redemption", as the apparent shield bubble could have actually been the bubble of the cloaking device.

Sight from inside a cloaked ship For all we know a cloaking device, for instance one of Romulan/Klingon flavor, renders a vessel invisible by bending any light around its hull. The effect is that neither active nor passive sensors would receive anything from the direction of the cloaked vessel, except for the star background (which we must assume is not distorted, because otherwise it would readily give away the ship's position, like hinted at in "Star Trek III"). But what about the sight from inside the ship? If it is surrounded by a perfectly opaque cloaking bubble, then the ship would not only be invisible but also blind. This is definitely not the case, as on all occasions without exception we could witness or assumed that the crew was able to observe the space around their cloaked ship.

A cloaked vessel could use specially modulated light to penetrate the cloaking, but then a scan across the spectrum could reveal its location, especially since this light would need to be emitted in all directions. Maybe some light from outside still reaches the hull in spite of the raised cloak, enough to be amplified to half-way clear images. However, such a "light trap" must be assumed to re-emit some radiation (most likely in the infrared range). An alternative and rather effortless explanation is that a tiny probe, which would not be visible on enemy scanners, is kept outside the cloaked region. Maybe this is the true reason for the periscope on the Klingon Bird-of-Prey in "Star Trek V" (although we definitely wouldn't expect a pipe to protrude from the hull of the vessel)?

Beaming with shields up It is an important concept, also for story-telling, that beaming is only possible when the shields are down. Entering the shield bubble with a shuttle isn't possible either, as we learned in "Star Trek V". In some episodes like TNG: "The Wounded" and TNG: "Relics", however, the transporter is definitely used while the shields are in place. In "The Wounded", O'Brien used a scan "window" in the shields of the U.S.S. Phoenix to beam over to the ship. In "Relics" we may assume that Scotty and Geordi, who were on the Jenolan to keep the door to the Dyson sphere open with the ship's shields, transmitted the shield frequency to the Enterprise. On the Enterprise, the frequency of the transporter carrier wave could be matched to that of the Jenolan shields. In a similar fashion the Duras sisters managed to penetrate the shields of the Enterprise-D in "Star Trek: Generations", after they had seen the frequency through Geordi's modified VISOR.

Matching phaser and shield frequencies In order to obstruct adaptation, the frequencies of ship and hand phasers are rotated in almost all of Starfleet's encounters with the Borg since TNG: "Best of Both Worlds". We may assume that the Borg analyze phaser shots and would be able to adjust their shields to block exactly the phaser frequency if this were not rotated. In other words, a Borg shield of a certain frequency would block or rather cancel out incoming fire of the same frequency through interference. In "Star Trek: Generations", however, quite the opposite happens. Here the Klingons, having found out the Enterprise's shield frequency through Geordi's VISOR, adjust their weapons to exactly this frequency to penetrate the shields. It seems that only Borg shields use the principle of interference, whereas Federation shields are designed rather to deflect, reflect or disperse incoming beams. Moreover, only waves of the same type would be able to interfere with each other. Considering that a phaser or disruptor is a matter wave, while the shields may be an electromagnetic field, a gravitational field (as the TNGTM states) or a subspace field, they should work on a different basis than interference. Maybe the shields are least efficient for some other reason at exactly the weapon's frequency.

There is one more problem. In "Generations" the Klingons match their phaser frequency with the Enterprise-D shields in order to penetrate them. But in order to fire through their own shields they would need to adapt these accordingly. In other words, they would open a window that may have allowed the Enterprise-D weapons to hit them likewise, but returning fire obviously failed while the Klingon disruptors were still firing. Either the different operation principle gives disruptors an advantage in mutual shield penetration, or shields work differently than we thought.

Tactics

Battle distances It is plausible that ships come very close to each other in a peaceful encounter, but the same seems to apply to battles. Although the distances in dialogues are usually hundreds or thousands of kilometers, their distance on screen is usually at most one kilometer. The real-world reason is obvious, as it is possible to shoot two ship miniatures at once if they are close to each other. Also for the drama it may be regarded useful if the enemies see each other almost "face to face" and not only on the viewscreen. It is controversial whether the definite statements or the definite visual evidence should be taken as true. I think that usually it would make no sense to come very close to a (possible) enemy, unless it was a suicide attack. The reaction times are simply too short, and the phaser yields can be considered still higher at short distances. Only for a small ship that is going to fight a large one, the close attack would be a useful tactic, as the large ship could make use only of a part of its weapons at once, and the small ship would be hard to hit at a short distance, like we have seen it in DS9: "Shattered Mirror" or DS9: "Valiant".