* Matthew Toseland <toad at amphibian.dyndns.org> [2008-06-02 15:03:37]:
> On Monday 02 June 2008 14:47, Florent Daigni?re wrote: > > * Matthew Toseland <toad at amphibian.dyndns.org> [2008-06-02 14:15:02]: > > > > > Okay, what issues do we have here? > > > > > > 1. The processes page takes longer than the copying files page. In fact, > most > > > of the time of the installer is occupied by the processes page. > > > > > > IMHO the solution to this is to bundle everything which is essential with > the > > > base installer. Then the processes page will take much less time. Also we > > > won't need a separate bundle-installer, so it will be easier to use in > places > > > where the Freenet website is blocked. The downside is the installer will > be > > > significantly bigger. > > > > I'm not convinced about that > > Why? It won't be huge, if we don't bundle any unnecessary Huge Stuff (see > below). And it's not small now. That means we will have to generate new installers on a regular basis... I don't like that :p > > > 2. The installer shouldn't download Thingamablog and Thaw. They are huge, > we > > > have not audited the code, and users can get them if they want them. > > > Debundling them would mean that as soon as we open the browser the > Processes > > > page will have completed, and the user can click Next and go on to > creating > > > icons. There seems to be a consensus on debundling, at least between Ian > and > > > Nextgens. :) I think a policy of not bundling anything we can't code > review > > > is reasonable, and clearly code reviewing Thingamablog for example would > be a > > > massive undertaking and isn't appropriate at this time. I do think we > should > > > keep the current plugins, however most of them are small. Even if we > include > > > the WoT plugin, that is also likely to be reasonably small for the > > > foreseeable future. > > > > Ok, let's get rid of them then > > For reasons of code reviewability? > Yep > Further to the above, I do actually review Thingamablog commits, however I > haven't reviewed the original, and it's not really widely enough used that we > can assume it to be safe? However, I do disagree with Ian's assertion that > nothing we don't own is reviewable: it is much easier to hide bugs in C > because you have buffer overflows, format string vulnerabilities and so on. > In Java, well written code *can* be reviewed, although of course more subtle > bugs may slip through the net. Also I review jSite commits (I dunno whether > the original on which the diffs are applied was reviewed though). > > > > > 3. Is it possible to change the names of the two panels? Copying Files vs > > > Setting Up Freenet, perhaps? > > > > We could get rid of one of the panel before the processing one > > Good idea. Will see what I can do > > > > > 4. Is it possible to open the browser after closing the installer rather > than > > > in the Processes page? > > > > Not "cleanly" but we could do a workaround time-delaying the browser > > startup. > > I'm not convinced that's better than the alternative. Okay, so WONTFIX -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20080602/8b9041c3/attachment.pgp>
