From: Duncan C. <dun...@wo...> - 2005-07-29 13:57:38
|
Axel, I was just wondering if there were any changes you knew of that we've got in cvs HEAD that we wouldn't want in the 0.9.8.1 release? I'm leaving out the child attributes (though event that is arguable as they work fine - opinions?). Are there any other changes to leave out? So I am including the GValue/GenericValue changes. I think they're ok but it could probably do with more testing, especially apps that use the tree/list stuff. Things still left in TODO: * audit to make sure that all enums that are used are exported * fix all FIXMEs in the documentation I'm only going to be semi-available this coming week. And the following week I am away. I'm in a play at the Edinburgh Fringe :-) http://www.twoshadesofblue.org.uk/twoshades.php?cast=display#company So if we release this coming week I might be able to build the windows installer. Otherwise by the time I'm back 6.4.1 will probably be out and I'd build the Windows binaries using that. A quick source-only release would still be good however to fix our most egregious bugs (we're currently blocking development of a couple apps). Duncan |
From: Axel S. <A....@ke...> - 2005-07-30 14:55:21
|
Dear Chief Weasel, On Fri, 2005-07-29 at 14:59 +0100, Duncan Coutts wrote: > Axel, > > I was just wondering if there were any changes you knew of that we've > got in cvs HEAD that we wouldn't want in the 0.9.8.1 release? Since we're (un)breaking things, there doesn't seem to be a big reason not to simply add new stuff. Is the segfault with disconnecting signals fixed? > I'm leaving out the child attributes (though event that is arguable as > they work fine - opinions?). Are there any other changes to leave out? Could you refresh my memory: what were these child attributes? > So I am including the GValue/GenericValue changes. I think they're ok > but it could probably do with more testing, especially apps that use the > tree/list stuff. > > Things still left in TODO: > > * audit to make sure that all enums that are used are exported Didn't you have a plan for this? > * fix all FIXMEs in the documentation Yes, I've seen FIXMEs in e.g. widget. I'd like to add a function called newPangoLayout which is simply missing and which I've been asked about. > I'm only going to be semi-available this coming week. And the following > week I am away. I'm in a play at the Edinburgh Fringe :-) > http://www.twoshadesofblue.org.uk/twoshades.php?cast=display#company > > So if we release this coming week I might be able to build the windows > installer. Otherwise by the time I'm back 6.4.1 will probably be out and > I'd build the Windows binaries using that. Ok, let's try to release a new source tar next week. Weather we delay the windows installer doesn't matter that much. Will the packaging stuff break with 6.4.1? Do we split the FIXME fixes? Which ones should I do? Axel. |
From: Duncan C. <dun...@wo...> - 2005-07-30 19:14:15
|
In message <112...@am...> Axel Simon <A....@ke...> writes: > Dear Chief Weasel, > > On Fri, 2005-07-29 at 14:59 +0100, Duncan Coutts wrote: > > Axel, > > > > I was just wondering if there were any changes you knew of that we've > > got in cvs HEAD that we wouldn't want in the 0.9.8.1 release? > > Since we're (un)breaking things, there doesn't seem to be a big reason > not to simply add new stuff. Yeah, I guess so. > Is the segfault with disconnecting signals fixed? Seems to be, but I don't understand why it happened in the first place. I can't see why control is re-entering Haskell land. > > I'm leaving out the child attributes (though event that is arguable as > > they work fine - opinions?). Are there any other changes to leave out? > > Could you refresh my memory: what were these child attributes? These are attributes that apply not to the object or to the child but to the relation of the two. Typical things are packing attributes of container children. It doesn't break anything to add them so it'd be ok if we're adding stuff. > > So I am including the GValue/GenericValue changes. I think they're ok > > but it could probably do with more testing, especially apps that use the > > tree/list stuff. > > > > Things still left in TODO: > > > > * audit to make sure that all enums that are used are exported > > Didn't you have a plan for this? Eventaully the code generator could deal with this but not yet. I don't think I had any other particular plan. > > * fix all FIXMEs in the documentation > > Yes, I've seen FIXMEs in e.g. widget. Yes a few slipped through. > I'd like to add a function called newPangoLayout which is simply missing > and which I've been asked about. Ok. > > I'm only going to be semi-available this coming week. And the following > > week I am away. I'm in a play at the Edinburgh Fringe :-) > > http://www.twoshadesofblue.org.uk/twoshades.php?cast=display#company > > > > So if we release this coming week I might be able to build the windows > > installer. Otherwise by the time I'm back 6.4.1 will probably be out and > > I'd build the Windows binaries using that. > > Ok, let's try to release a new source tar next week. Weather we delay > the windows installer doesn't matter that much. Will the packaging stuff > break with 6.4.1? Hopefully not. 6.4.1 should contain a great many bug fixes but I don't expect any breakage. > Do we split the FIXME fixes? Which ones should I do? Not sure yet, I might have some time tomorrow evening. I've only got email access today. Duncan |
From: Duncan C. <dun...@wo...> - 2005-08-01 00:59:08
|
In message <112...@am...> Axel Simon <A....@ke...> writes: > Dear Chief Weasel, Ha! I didn't notice that the first time I replied. :-) > > Things still left in TODO: > > > > * audit to make sure that all enums that are used are exported > > Didn't you have a plan for this? Ok so my plan goes like this: copy the Enum.hs module(s) and transform them into modules that use every enum type name and value. This will test that they are all exported. I've done that for the gtk enums module so far, the gdk enums module is still to go. Result, several enums are never used at all and several others are now exported from the apprpriate modules. The SF cvs seems to be down just at the moment. I'll try again tomorrow. Seeing as we seem to be including all our chages since 0.9.8 then we should make the "branch-0-9-8-1" branch from MAIN rather than taggin another release from the "branch-0-9-8". So I'll not bother merging anything to that branch. Duncan |
From: Axel S. <A....@ke...> - 2005-08-01 08:39:59
|
On Mon, 2005-08-01 at 01:59 +0100, Duncan Coutts wrote: > In message <112...@am...> Axel Simon > <A....@ke...> writes: > > Dear Chief Weasel, > > Ha! I didn't notice that the first time I replied. :-) > > > > Things still left in TODO: > > > > > > * audit to make sure that all enums that are used are exported > > > > Didn't you have a plan for this? > > Ok so my plan goes like this: copy the Enum.hs module(s) and transform them into > modules that use every enum type name and value. This will test that they are > all exported. Ok, sounds like a plan. > Seeing as we seem to be including all our chages since 0.9.8 then we should make > the "branch-0-9-8-1" branch from MAIN rather than taggin another release from > the "branch-0-9-8". So I'll not bother merging anything to that branch. Yes. I suppose, to save work, you shouldn't branch before you actually say 'make dist'. It's not worth the hassle, since there is hardly any development work going on. Having said that, I spent the last two days finishing off Pango. genApi isn't particularly useful, I suppose Mono doesn't cover all there is to cover in Pango. But it's not that much, it's just tricky (e.g. converting UTF8 byte indices to String indices). I'm far from finished, but I can't commit until we make the source ball release anyway. Axel. |