Closed
Bug 166007
Opened 22 years ago
Closed 22 years ago
Fix formatting in GIF2.cpp
Categories
(Core :: Graphics: ImageLib, defect)
Core
Graphics: ImageLib
Tracking
()
RESOLVED
FIXED
mozilla1.3alpha
People
(Reporter: paper, Assigned: paper)
Details
Attachments
(2 files, 4 obsolete files)
76.78 KB,
patch
|
Biesinger
:
review+
tor
:
superreview+
|
Details | Diff | Splinter Review |
33.23 KB,
text/plain
|
Details |
Making patches for GIF2.cpp bugs me to no end because it's abnormally set to tab-width 4. tab-width should be 2.
Assignee | ||
Updated•22 years ago
|
Status: UNCONFIRMED → ASSIGNED
Ever confirmed: true
Assignee | ||
Updated•22 years ago
|
Summary: Change GIF2.cpp to tab-width 2 → Fix formatting in GIF2.cpp
Target Milestone: --- → mozilla1.3alpha
Assignee | ||
Comment 1•22 years ago
|
||
- 0 code changes. - 4 space tabs changed to 2 space tabs - Removed all ILTRACEs (they were all commented out and were for old gifcom stuff) - Removed debug defines (again, only used way back in gifcom days) - Placed return value type & function name on same line (ie "void^pfoo" to "void foo()") - moved { to same line as if, and "} else {" - Removed trailing spaces - Split any lines > 80 characters. All this chopped 200+ lines off the file and made it much more consistent and standardized to the rest of imglib. :)
Comment 2•22 years ago
|
||
Comment on attachment 104248 [details] [diff] [review] Cleanup can you attach a diff -w?
Assignee | ||
Comment 3•22 years ago
|
||
still rather long, but easier to find out what I did.
Comment 4•22 years ago
|
||
Comment on attachment 104280 [details]
diff -u2 -w
what's this?
+ // q[6] = Pixel Aspect Ratio
+ // Not used
+ // float aspect = (float)((q[6] + 15) / 64.0);
why add this code?
Assignee | ||
Comment 5•22 years ago
|
||
The code below wasn't doing anything, so I killed it, but I thought the information in it might be usefull in the future. (ie. someone asking why we skipping over [6] or something like that) - if (q[6]) - { - /* should assert gif89 */ - if (q[6] != 49) - { -#ifdef DEBUG - float aspect = (float)((q[6] + 15) / 64.0); - //ILTRACE(2, ("il:gif: %f aspect ratio", aspect)); -#endif - } - }
Comment 6•22 years ago
|
||
+ if ( !gs->prefix || !gs->suffix || !gs->stack) { ^ remove the space here please (other places like this as well) + if (strncmp((char*)q,"GIF",3)) { space after comma, please same for this one (which you didn't touch): GETN(7,gif_global_header); and possibly in other places of that file. -#ifdef DEBUG_saari given the amount of checkins saari has done recently, I suppose that's ok. + if (*q!=',') { space before and after != please. + for (int i=0; i < gs->local_colormap_size; i++, map++) { space around = please + // XXX: we don't freeze decoder anymore um, is the non-freezing a bad thing? doesn't sound like one to me; and if it's good, it should not be XXX ok the rest looks good, but I still have to apply it and test it.
Comment 7•22 years ago
|
||
ok I now applied the patch and looked at the resulting file. I found a few things, I hope you don't mind :) : /* PR_ASSERT(0); */ you probably should remove that comment (or replace with an NS_NOTREACHED) } while(gs->irow > gs->height - 1); personally, I'd like parens around the second half of the comparison... #define OUTPUT_ROW(gs) this macro could use NSPR_BEGIN_MACRO / NSPR_END_MACRO for (ch=q; count-- > 0; ch++) spaces around = gif_struct *ret; ret = PR_NEWZAP(gif_struct); combine these lines? actually, maybe assign it directly into the out parameter? (or does code depend on an unchanged parameter in case it would be zero?) hm, not sure if this really belongs in this bug... PR_ASSERT(gs); can you replace that with NS_ASSERTION(gs, "Got null argument, will crash!"); GIF_IRGB *img_trans_pixel; I'd move it inside the if(), to where it's first used. ...gotta go now, will continue later. for my own reference, I'm near line 561 now.
Comment 8•22 years ago
|
||
if (gs->gathered < max) that if condition can be written as return (gs->gathered < max); PRStatus gif_write(gif_struct *gs, const PRUint8 *buf, PRUint32 len) well I wish that would use nsresult but that does not really belong in this bug. // PR_ASSERT ((len == 0) || (gs->gathered < MAX_READ_AHEAD)); guess you can remove this comment const PRUint8 *q, *p=buf, *ep=buf+len; again, spaces around =. I'll not mention further cases of this, just fix it everywhere. if (!((len == 0) || (gs->gathered < MAX_READ_AHEAD))) now this line looks, uh, rather ugly. maybe change it to: if ((len != 0) && (gs->gathered >= MAX_READ_AHEAD)) (wow, I really hope I got the boolean algebra correct here...) actually, for all the *(q + 8), you could write q[8] instead. that's better, imho. if ((num_colors > gs->local_colormap_size) && gs->local_colormap) { PR_FREEIF(gs->local_colormap); gs->local_colormap = NULL; } now this piece of code is interesting (as mentioned on irc) in short, change the FREEIF to PR_Free, because FREEIF would also check for non-nullness of local_colormap, and set it to NULL afterwards. alternatively, remove the non-nullcheck from the if and the assignment of null. now I see the NULL in this code... please do a global search-and-replace for NULL -> nsnull. GIF_RGB* map; map = gs->local_colormap; combine these two lines please. return PR_FAILURE; break; no need for the break;s here. also in surrounding code. ah yeah, and here (at the end of the file): if (gs->local_colormap) { PR_FREEIF(gs->local_colormap); gs->local_colormap = NULL; } you can also replace all these lines with the PR_FREEIF or change the PR_FREEIF to PR_Free great, I'm done.
Assignee | ||
Comment 9•22 years ago
|
||
Attachment #104248 -
Attachment is obsolete: true
Assignee | ||
Comment 10•22 years ago
|
||
> + // XXX: we don't freeze decoder anymore I removed that whole chunk. We haven't frozen loading GIF in Mozilla since.. well, since the time before Mozilla was public. > + if (strncmp((char*)q,"GIF",3)) { > space after comma, please > same for this one (which you didn't touch): > GETN(7,gif_global_header); Done, plus all similar > /* PR_ASSERT(0); */ chopped :) > } while(gs->irow > gs->height - 1); > personally, I'd like parens around the second half of the comparison... Added parens (didn't remove old ones, they seem to be standard on while). Also added a space between while and (, and did that to other areas of code with a while(, switch(, or return(. > #define OUTPUT_ROW(gs) > this macro could use NSPR_BEGIN_MACRO / NSPR_END_MACRO Added PR_BEGIN/END_MACRO to this one, and removed NS from the other one. > ret = PR_NEWZAP(gif_struct); I'm going to leave this as is. It does seem constructed a bit weird though. > GIF_IRGB *img_trans_pixel; >I'd move it inside the if(), to where it's first used. That whole function (gif_init_transparency) is a mess and could probably be cut in half if it weren't for my theory that the function serves no purpose anymore. I've made a note to look deeper into my theory, and will leave the code as is here. I kept the PR_FREEIF, and removed the null check prior to it and the setting to null after it. Anything else I didn't mention, I changed as per comments.
Attachment #104280 -
Attachment is obsolete: true
Assignee | ||
Comment 11•22 years ago
|
||
more clean-up as per discussion with biesi on IRC
Attachment #105616 -
Attachment is obsolete: true
Assignee | ||
Comment 12•22 years ago
|
||
Attachment #105617 -
Attachment is obsolete: true
Comment 13•22 years ago
|
||
Comment on attachment 105625 [details] [diff] [review] Super Clean-Up r=biesi
Attachment #105625 -
Flags: review+
Comment 14•22 years ago
|
||
Comment on attachment 105625 [details] [diff] [review] Super Clean-Up Somebody set us up the bomb. All your GIF2.cpp are belong to Paper! sr=tor
Attachment #105625 -
Flags: superreview+
Assignee | ||
Comment 15•22 years ago
|
||
Checking in GIF2.cpp; /cvsroot/mozilla/modules/libpr0n/decoders/gif/GIF2.cpp,v <-- GIF2.cpp new revision: 1.35; previous revision: 1.34 done Marking Fixed
Status: ASSIGNED → RESOLVED
Closed: 22 years ago
Resolution: --- → FIXED
You need to log in
before you can comment on or make changes to this bug.
Description
•