John J Foerch <[EMAIL PROTECTED]> writes: > Jeremy Maitin-Shepard <[EMAIL PROTECTED]> writes: >> I think that for many sites, a page mode can be designed such that it is >> essentially a strict improvement, in that everyone, or nearly everyone, >> prefers it. For media sites like youtube and google video where the >> mode merely specifies how to find the embedded media file and changes >> some operations like save (s) and shell-command-on-file (x) and >> shell-commmand-on-url (X) to operate on the embedded media, it is hard >> to imagine that anyone would find it undesirable, and I think nearly all >> users would prefer that as many such modes as possible are loaded by >> default.
> I think I basically agree with this. However, it seems very likely > that before long we will have lots of site-specific code that will be in > a category of fantastic for the experienced user, but frustrating for > the newbie. No matter where I am on the web, I want to be able to use > the same keys in the browser to do the same stuff--not have to pause to > learn more exceptional behavior of the browser. On that note, I would > even prefer that page-modes not change my default hint classes. I agree that it is best not to make unsolicited changes, but it seems in the case of page modes for "media" sites, the unchanged default hint classes for save, shell-command-on-file, and shell-command-on-url are almost certainly not desirable. >> I don't think saving disk space is really an important argument at all, >> as Mozilla is very large itself. The page modes are often quite small, >> so even hundreds or thousands of them wouldn't take up that much space, >> though I suppose if it really got to that point it may well make sense >> to somehow separate them into a different repository or coming up with >> some better distribution method. > Note I wasn't referring to disk space, but overall compactness, in the > sense that a smaller program is easier to understand and to hack. I > would expect that given the instability of the mozilla platform, and our > own evolving ideas of what conkeror should be, the fewer dependencies > that have to be dealt with to make core changes, the better. If we get > to the point where we are packaging something that provides M-x doctor, > we will know we have gone too far. ;) Well, pushing page specific modes out as "external packages" doesn't really solve the problem of having to fix them as changes in other parts of Conkeror break them; rather, it merely attempts to pretend that it is someone else's problem (but in fact it would probably still be our problem, since we may well have written many of them), and even so, the goal is not to encourage page modes to bitrot. In fact, having them all packaged in Conkeror makes it much easier to just do a search for uses of a particular function than if they are scattered around in multiple locations, or if the person making the change doesn't even have any real way to know what external modules may be using the changed interface. -- Jeremy Maitin-Shepard _______________________________________________ Conkeror mailing list [email protected] https://www.mozdev.org/mailman/listinfo/conkeror
