January 2013 - Dillo-dev - lists.mailman3.com

chain.c cleanup
by p37sitdu＠lavabit.com 02 Feb '13

02 Feb '13

http://www.inventati.org/noname/dillo/chain-cleanup.patch 1. Replaced dNew with dMalloc and dNew0 with dMalloc0. This way it is possible to use (sizeof *var) instead of repeating variable type. By the way, when 'count' is constant, dMalloc is easier to use, and when count is not constant, integer overflow can happen in dNew during multiplication. It looks like there are no such errors, but I think check should be added or, better, dCalloc can be introduced so libc checks for such cases. 2. Replaced #defines with enums and made them uppercase so it is easier to distinguish them from variables. This style is already used in most modules. IO should be fixed too, probably. 3. Moved duplicate code from a_Chain_fcb and a_Chain_bcb to Chain_cb. Anonymous structures for connections (forward or backward) are introduced in ChainLink structure for this purpose. 4. Removed unused a_Chain_unlink function. Chain links are just dFreed instead. Anyway, there is no function to connect existing links back, so disconnected link can only be dFreed. 5. Replaced AtoB_branch and BtoA_branch parameters of a_Chain_link_new with single Branch parameter. They were always equal anyway, and using different branch identifiers for one connection makes no sense, as branch ID is used to distinguish between Answer and Query chains. Branch parameter can be eliminated completely by separating existing CCC callbacks into answer and query callbacks. Currently instead of two functions we have one with if (Branch == 1) { ... } else { ... }

1 1

Yet another SGI compiler error...
by csoren＠isd.net 31 Jan '13

31 Jan '13

> > > > What's the invocation for checking out the latest commit > > from the repository? I didn't see it in the developer > > section of the website. I assume `hg clone dillo-3` > > grabs 3.0.2 as it existed on it's release date? > > > > "hg clone dillo3" will simply copy *your* repository, not > the one from hg.dillo.org. Cloning again the repository > from hg.dillo.org will indeed get you the newest changes: > > mv dillo3 dillo3_copy > hg clone http://hg.dillo.org/dillo dillo3 > cd dillo3 > > But this is faster: in the repository (after "cd dillo3"), > type > > hg pull http://hg.dillo.org/dillo > hg up > > to get the newest changes. "hg pull", without the URL, > should be sufficient in most cases. > > HTH. > > Sebastian > Okay thanks! Sorry for the newb questions; I've never used mercurial before... 0_o

1 0

Yet another SGI compiler error...
by csoren＠isd.net 31 Jan '13

31 Jan '13

> On Wed, Jan 30, Chris Sorenson wrote: > > > > > > Thoughts? > > > > Try the latest commit. > > Sebastian > What's the invocation for checking out the latest commit from the repository? I didn't see it in the developer section of the website. I assume `hg clone dillo-3` grabs 3.0.2 as it existed on it's release date?

2 1

Yet another SGI compiler error...
by csoren＠isd.net 30 Jan '13

30 Jan '13

Hi, I've gotten 27 files to compile so far! But then this happened: <pre> -------------------------------------- cc-1238 CC: ERROR File = table.cc, Line = 880 The member "dw::core::Widget::extremes" is inaccessible. col,extremes->get(col).minWidth, colExtremes->get(col).maxWidth); ^ cc-1044 CC: ERROR File = table.cc, Line = 880 The indicated expression must have pointer type. col,extremes->get(col).minWidth, colExtremes->get(col).maxWidth) -------------------------------------- </pre> Thoughts?

2 1

CustChoice
by corvid＠lavabit.com 28 Jan '13

28 Jan '13

I was looking into why the search dialog would no longer let me tab to the search engine list and press down arrow to switch between them... It seems that, once Jorge added a CustChoice in fltkui.cc, the dialog code started to use _that_ one and not its own. (I would've expected something to complain about there being two CustChoice::handle()s in that case...) Would it be considered incorrect in any way if fltkui.cc's CustChoice and CustInput2 were put inside dw::fltk::ui?

2 5

Another SGI compiler problem
by csoren＠isd.net 27 Jan '13

27 Jan '13

> > Any problem if you change > bool FltkToggleButtonResource::FltkToggleButtonResource::isActivated () > to > bool FltkToggleButtonResource::isActivated () > ? > Thanks that worked!

1 0

Another SGi compiler problem
by csoren＠isd.net 24 Jan '13

24 Jan '13

> > > > Greets all, > > > > I'm trying to compile on Silicon Graphics (IRIX O.S.) > > using their MIPSPro 7.4 > > compiler, but getting this error in dialog.cc: > > > > <pre> > > cc-1028 CC: ERROR File = dialog.cc, Line = 180 > > The expression used must have a constant value. > > > > memset(pm, '\0', sizeof(Fl_Menu_Item[n_it+1])); > > ^ > > </pre> > > > > Thoughts? > > You may try something like: > > memset(pm, '\0', (n_it+1)*sizeof(Fl_Menu_Item)); > > -- > Cheers > Jorge.- Thanks Jorge that worked! :) Fourteen files compiled so far but then this happened in fltkflatview.cc: ---------------------------------------------- cc-3168 CC: ERROR File = fltkui.cc, Line = 782 A name followed by "::" must be a class or namespace name. bool FltkToggleButtonResource::FltkToggleButtonResource::isActivated () ^ cc-1443 CC: ERROR File = fltkui.cc, Line = 782 "dw::fltk::ui::FltkToggleButtonResource::FltkToggleButtonResource" is not a class template. bool FltkToggleButtonResource::FltkToggleButtonResource::isActivated () ^ 2 errors detected in the compilation of "fltkflatview.cc". ----------------------------------------------- Thoughts?

2 1

[patch] dCalloc
by p37sitdu＠lavabit.com 24 Jan '13

24 Jan '13

dNew macro hides multiplication and can lead to integer overflows. Patch implements dCalloc and replaces dNew with dCalloc wrapper so no integer overflows are possible: http://www.inventati.org/noname/dillo/dlib/dcalloc.patch dwrite.patch is moved to http://www.inventati.org/noname/dillo/dlib/

1 0

_t suffix
by p37sitdu＠lavabit.com 23 Jan '13

23 Jan '13

Suffix _t is reverved by POSIX and should not be used: <http://pubs.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html> I have made patches that remove _t suffix from some modules: http://www.inventati.org/noname/dillo/_t/

1 0

line wrap for Japanese
by furaisanjin＠gmail.com 23 Jan '13

23 Jan '13

Hello all. It seems that change set de05508cbbbd causes the line wrap problem for Japanese. Certain lines become very long and I don't see appropriate line wrap in the middle. Regards, furaisanjin

2 2