-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hi Ralf, while I'm not doing MIDI in any way for myself, I am watching this threads with some general interest since some time. But I don't get any clue what is the actual/real/precise/technical problem that plagues you. At this point I should give some background information to make my question clear. I do CD recordings for classical music with the aim of creating quality oriented recordings. Which means, for example, that I have the necessary equipment and training to notice when a spot mike goes away from the "optimal" position I settled down for this mix. Even just for such a small amount as of 1 or two sample steps. 1/96000 s = 10 micro seconds. Of course you can't hear such as a timing difference, but in the context I am working, you can hear it as a colouring and unevenness in the way the spot supports the main micro. I wouldn't call this effect minute or esoteric, it's quite blatant, at least for trained ears, /and given the specific situation/ I am working in. Working with Ardour in this context can be painful, to be frank. I am still using Ardour 0.99 -- but this doesn't make any difference for my argument, because 0.99 can't be considered an early alpha version. Rather, it has to be assessed as an almost complete first version of an professional DAW. (Btw. I can't just switch to a current or new version of Ardour in the middle of projects, because of the risk and because of the numerous additional python scripts I created to get my work done) Now, to get to my point. You frequently can find me quite angry at "those hacker guys, which apparently have wood chips instead of ears". Imagine, an allegedly "professional" application which is plagued by off-by-one errors in the snap functions, imprecisions with the waveform display, partially unable to re-import the mixdown of a range with absolute precision or to make click-free splits. Which can't record 96/24 uninterrupted for longer than 1.5 hours and which can't handle timelines longer than about 5-8 hours. Not to mention the general "uneasyness" of handling. And yes, it crashes quite frequently (hopefully 2.x is improved with this respect) But now, when reconsidering, the situation isn't the way it may look. First, I should add that I am myself a professional and open source developer and know the work situation of "those hacker guys" inside out. I know, what it means e.g. that GTK 1 just wasn't up to the job but sadly switching to GTK 2 and learning form errors actually incurs a partial redesign and rewrite of the codebase -- thus being thrown back in the development by months or even years. The next interesting points to note is that most of the "blatant" problems are already reported as bugs in Mantis since a long time. And the general uneasyness of the UI evaporates into nothing the moment you switch the working and music style. For example, recording vocals with punch on a fixed bars/beats grid in a studio like setup can be quite fun with ardour. Suddenly, the same UI feels smooth and well suited for the job, and the app is rock stable. While, on the surface, programming work has much to do with logic and things being 100% exactly defined, actually there is a second level of meaning right behind the facade. Each software development project has a more or less explicit image of what the intended use of the software might be, i.e. the way the prospective user feels and behaves. And to my own experience, it is notorious for the programmer to partially perceive the real world user as someone doing wild and wired things to "his" application (while actually the reason is just a mismatch on the mentioned "image of intended use"). Which puts the user in the strange and often impossible situation of having to prove something is broken. Ralf, you describe a reasonable concrete usage situation. I take it that you have checked the bulk of the available sequences for use in this specific situation. > Edit by using a grid or not editor C, Am, F, G. After that copy and > paste it, until it has a length of 2 minutes. Do a loop range of some > periods of C, Am, F, G, maybe 2 or 4 times. Let the loop range play > and on a second MIDI track play for another instrument any white key > on the manual, you can ride on C, Am, F, G by playing the white keys. > Do this without recording for about 10 minutes, than record. Repeat > this for the whole song, step by step and while doing that, change > the soundfonts. > And now, what is the problem? Ralf Mardorf schrieb: > I can't do that with the most stable Linux I've got. Any Linux of > mine can play MIDI files, but I can't produce with any of my Linux > MIDI music myself. And this means...? (a) all Seqs you tested behave unstable in this situation and crash? (b) -or- you get latency problems when playing live? (c) -or- the timings are unreliable and not exactly reproducible? (d) -or- you have problems when coupling multiple apps in sync? (e) -or- you have problems getting your audio hardware working? Or maybe all of those problems? And, further, apparently there is one breed of users which seem to be quite fine, while you mention others (friends, complaints on forums) which seem to experience similar fundamental problems. Any chance to identify a key difference between those two user groups? And, moreover, does anyone here know if the upstream developers are aware of the problem. If not, do we have any chance of making them aware -- I mean in a way they can't just put it off as being a strange personal opinion? Cheers, Hermann Vosseler -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJJ1vaZbZrB6HelLIRAguyAKCFoRllAH8a7z3Ian6ORAp7HzKu1gCg6I+W oci6OSWlEGTD4gRcMoG6yUQ= =dMUE -----END PGP SIGNATURE----- _______________________________________________ 64studio-users mailing list [email protected] http://lists.64studio.com/mailman/listinfo/64studio-users
