Code::Blocks Forums
User forums => Nightly builds => Topic started by: killerbot on February 18, 2008, 09:46:11 pm
-
Get quick announcements through the RSS feed http://www.codeblocks.org/nightly/CodeBlock_RSS.xml
Before you use a nightly make sure you understand how it works (http://forums.codeblocks.org/index.php/topic,3232.0.html).
A link to the unicode windows wxWidget dll for Code::Blocks : http://prdownload.berlios.de/codeblocks/wxmsw28u_gcc_cb_wx287.7z
For those who might need this one (when no MingW installed on your system) : the mingw10m.dll : http://prdownload.berlios.de/codeblocks/mingwm10_gcc421.7z
The 18 February 2008 build is out.
- Windows :
http://prdownload.berlios.de/codeblocks/CB_20080218_rev4893_win32.7z
- Linux :
none
Resolved Fixed:
- Minor fixes in debugger (disassembly cursor tracking, etc)
- Fixed: Erratic behaviour (not all opened editor were shown with their relative path) of Opened files tree right after a Project load
- added BrowseTracker to contribs
- help_plugin: add help files of index.ini only once
- compiler autodetection: allow mingw compiler suite to be installed under [C::B]\mingw, too
- Help plugin: Help files read from index.ini are now added in *bold* to the help menu, cannot be edited through the configuration dialog and are not saved to the Code::Blocks config file
Regressions/Confirmed/Annoying/Common bugs:
- toolbar-images-not-changing-state (is a wx problem/Win XP problem)
-
sounds like the number of bug fixes and frequency is going down a lot, does this mean we are near a 1.0f?
-
Build for Ubuntu i386/amd64 can be found here (http://lgp203.free.fr/spip/spip.php?article1)
Best regards, pasgui
-
Ubuntu 7.04 -> 7.10 Amd64 tar.gz archive (containing '.deb' installers builds with wx2.8.7) can be found here[
(http://www.esnips.com/web/CodeBlocks)
-
- Mac OS X:
http://prdownload.berlios.de/codeblocks/CB_20080218_rev4893_mac287.zip
-
Binaries for debian i386, amd64 and sources (tar-balls also usable on other platforms) can be found in my repository (see signature).
-
sounds like the number of bug fixes and frequency is going down a lot, does this mean we are near a 1.0f?
I hope not. Take a look at the bug tracker list. It's huge. And most of the bugs are old. Not to mention all the codecompletion bugs. I've had a hard time tracking down and reproducing them, but for the most part, I can tell that the codecompletion doesn't like default arguments. Neither does the symbol browser.
Ideally, when version goes 1.0f, you should be able to compile not only C::B, but also mingw with all the features being used without any bugs.
IMHO, of course.
-
And most of the bugs are old. Not to mention all the codecompletion bugs. I've had a hard time tracking down and reproducing them, but for the most part, I can tell that the codecompletion doesn't like default arguments. Neither does the symbol browser.
HHHhmmm. As developers we have chosen and committed to C::B as our development environment. We found it to be a mature platform with code completion and debugging working solidly. Ask someone who is using it hands-on (in this case SUSE 10.2 w/ C++) and you will get a big THANK YOU to all the contributors who make C::B what it is today. To us it makes no difference whether there are nightly updates or major releases. What matters is it works and is gettting better every day!
-
What matters is it works and is gettting better every day!
Thats the point :wink:
-
I have a problem: when I open my project by double-click the .cbp file, I can't use "Find declaration of:" or "Find implementaion of:" in the right-click menu; but when I do it with "File"->"Open", all works well.
Is it a bug !?
-
I have a problem: when I open my project by double-click the .cbp file, I can't use "Find declaration of:" or "Find implementaion of:" in the right-click menu; but when I do it with "File"->"Open", all works well.
Is it a bug !?
I have the same problem (as I stated here (http://forums.codeblocks.org/index.php/topic,7381.msg58791.html#msg58791)) and yes I think it's a bug in the CodeCompletion.
-
sounds like the number of bug fixes and frequency is going down a lot, does this mean we are near a 1.0f?
I hope not. Take a look at the bug tracker list. It's huge. And most of the bugs are old. Not to mention all the codecompletion bugs. I've had a hard time tracking down and reproducing them, but for the most part, I can tell that the codecompletion doesn't like default arguments. Neither does the symbol browser.
As a former developer of the codecompletion module, I can tell you that code completion WON'T be fixed until this revision is out. The code's too messy to be modified in a non-release branch. Many things need to be changed and updated.
As for the bugs in the tracker, I think it'd be better to start up with a fresh buglist. Most of the bugs there should be closed anyway.
-
As for the bugs in the tracker, I think it'd be better to start up with a fresh buglist. Most of the bugs there should be closed anyway.
Excuse me, but that's a bit weird. Most the people who wrote to this bug list had the view, that someone would act with the announced errors. That sounds like you have lost the track. Simply close the list and forget all that can't be the way of choice. I'm afraid that some important information from work from users would be thrown in the bin. So, isn't it better to close the solved and useless items?
And an note to "release or not": I'm working with different IDE's from Borland and Microsoft. They are expensive AND have a lot of bugs. Every new version offers new and more horrible problems as I can say. Codeblocks though is very stable in daily use, nearly every nightly could be labeled as stable in comparison. But there you go, I'm happy with nightlys or perhaps weeklys :lol:
Codeblocks is great, good work!
-
Excuse me, but that's a bit weird. Most the people who wrote to this bug list had the view, that someone would act with the announced errors. That sounds like you have lost the track. Simply close the list and forget all that can't be the way of choice.
It was said already two years ago that the bugs in code completion would not be fixed. It was repeated many times that code completion is considered pretty much FUBAR from the bottom and that we almost certainly won't address the well-known issues in this implementation.
People still keep adding bug reports (many of which refer to the exact same issues) even today. What is this telling you?
-
As for the bugs in the tracker, I think it'd be better to start up with a fresh buglist. Most of the bugs there should be closed anyway.
And I hope that C::B project doesn't continue to use the Berlios' bug tracker. It just doesn't have any of the basic features you would expect from a bug tracker.
-
Berlios offers the choice of using a Mantis issue tracker, which is a decent choice.
-
Berlios offers the choice of using a Mantis issue tracker, which is a decent choice.
Agreed. Looks much better.
-
I have two NOT codecompletion related issues =)
They rank about the custom compilation commands defined via the advanced compiler options.
I've setup a command that transforms html files to header files to include them as a resource. The first problem is that these files are always "compiled", although I modified the program that does the conversion to equalize the timestamps (ie the output has the same modification time as the source). The second problem is that IF they need recompilation, depending cpp files don't get recompiled in the same run, so basically CB doesn't reevaluate dependencies when compilation runs. I see that this is a vaild optimization, but for custom commands it's an error source. CB should recognize that a command creates c/h/cpp/hpp files (or more generally "files that are sources for other compilation commands") and reevaluate compilation necessity after they are "compiled" or, as an alternative, have a switch for custom commands that says "reevaluate dependencies after running this".
That problem also influences compilation order, but that can be fixed by assigning priorities.
-
codeblocks crashes on start :(
i tried first a selfcompiled version from svn Revision: 4899 then i tried the debs from jens.
i am on debian, amd64.
crash log is attached
--Ano
[attachment deleted by admin]
-
Agreed. Looks much better.
Pretty slow though. I have set up the same tracker at work due to each simplicity (it was a requirement) and it is really fast. I think that the setup is not optimal (they have all the berlios projects under one instance or something I guess). Anyway it is a full featured tracker, far better than the totally unacceptable simple berlios issue tracker.
-
There are two main problems in projects development: The First - to start, The Second - to accomplish :lol:. I have 38 issues in my project and nobody in my team nose how to close them :shock:.
-
When I do this sequence of steps: "Plugins | Class wizard" and I press the "Create button" (I didn't write anything on the dialog fields), CB crashes...
Is somebody else experiencing this behavior?
I am using svn4893 (Jens build) on Ubuntu 7.10 i386.
Thanks a lot! friends
-
When I do this sequence of steps: "Plugins | Class wizard" and I press the "Create button" (I didn't write anything on the dialog fields), CB crashes...
Is somebody else experiencing this behavior?
I am using svn4893 (Jens build) on Ubuntu 7.10 i386.
Thanks a lot! friends
Yes. Confirmed on Windows with 4893.
-
Yes. Confirmed on Windows with 4893.
Fixed in revision 4903.
-
Let me bring your attention to a bug: In Linux (Fedora or Ubuntu), if you set your system locale to Chinese, gcc will output all warning and error messages in Chinese, which means a warning message will not contain the string "Warning". And Code::Blocks exactly scans gcc messages for the string "Warning" to count the number of warnings. And in this case, all warnings are counted as errors. The solution is to add a "LC_ALL=en_US.UTF-8" as a global environment variable for Code::Blocks.
-
Let me bring your attention to a bug: In Linux (Fedora or Ubuntu), if you set your system locale to Chinese, gcc will output all warning and error messages in Chinese, which means a warning message will not contain the string "Warning". And Code::Blocks exactly scans gcc messages for the string "Warning" to count the number of warnings. And in this case, all warnings are counted as errors. The solution is to add a "LC_ALL=en_US.UTF-8" as a global environment variable for Code::Blocks.
https://developer.berlios.de/bugs/?func=detailbug&group_id=5358&bug_id=11611
I found it half a year ago.
I think the redesign of compiler framework will fix it.
Now Codeblocks uses regex to parse the output of gcc.
Another annoying bug caused by the current framework is bug 11243 (https://developer.berlios.de/bugs/?func=detailbug&bug_id=11243&group_id=5358) (same as bug 11423 (https://developer.berlios.de/bugs/?func=detailbug&bug_id=11423&group_id=5358)).
Why can't there be any Chinese characters(or other non-english) in source file path? :D
Can't wait any longer for the new release 8.02 :wink:
-
I think I found a bug but I don't know if I should submit it here. Anyway, here it goes: Let's assume you build something and gives you errors/warnings. In the "messages panel" the "build messages" tab gets the focus. If you do a right click in the "build messages" tab a menu comes up. Until now everything works as it is intended. But if you select something in the menu(left click) the menu comes up again!!! And you have to click again in order to work. This is specific to the "build messages" tab the other tabs work fine.
I am using pasgui's build for Ubuntu 7.10 amd64
-
Aren't syntax errors supposed to show up with a little box on the left of the line that contains the error? It doesn't seem to occur with this build.
Edit: I just noticed that it DOES work correctly if you use a project, but does not work if you compile a single .cpp file.
Perhaps that is the intended behaviour.
Since I insist my students use projects (although of course they don't :) ) I do not find this to be a bug.
-
Fixed in revision 4903.
Thanks, Biplab. I want to take advantage of this reply for making other suggestion.
I think it would help to save time whether the class wizard had an option for automatically generating the Copy Constructor and the Assignment Operator for the new class. Normally I make them private for prevention (only in the header file).
Thanks again for your kindness.
-
1) In Edit mode Auto-complete (Ctrl-J) does not work
2) In Edit mode Complete code (Ctrl-SPACE) does not work
I am using Code:Blocks 4893 build on Windows XP.
/badboy
-
I has a same problem in here:
http://forums.codeblocks.org/index.php/topic,6702.0.html
and C::B version is:
svn build rev 4893 (2008-02-18 20:06:04) gcc 4.2.1 Windows/unicode
-
Could someone tell me why I'm getting "..."s in the margin of my editor? It's to the right of the line number, where breakpoints are shown.
See image below.
Thanks, guys.
[attachment deleted by admin]
-
Could someone tell me why I'm getting "..."s in the margin of my editor? It's to the right of the line number, where breakpoints are shown.
I believe this is due to the BrowseTracker plugin. Disable it and the dots are gone. (Alternatively you can use it of course.. .;-)).
-
I think it would help to save time whether the class wizard had an option for automatically generating the Copy Constructor and the Assignment Operator for the new class. Normally I make them private for prevention (only in the header file).
Please file a Feature Request at the following link so that this doesn't get lost. :)
https://developer.berlios.de/feature/?group_id=5358
-
Please file a Feature Request at the following link so that this doesn't get lost. :)
You don't need it. It's implemented on my machine already. However - commit will have to wait a bit...
-
Please file a Feature Request at the following link so that this doesn't get lost. :)
You don't need it. It's implemented on my machine already. However - commit will have to wait a bit...
Thank you =-)
-
Please file a Feature Request at the following link so that this doesn't get lost. :)
You don't need it. It's implemented on my machine already. However - commit will have to wait a bit...
Thanks, Biplab and Morten. I love CB is getting better day after day :D
I believe I have other ideas to contribute in the Berlios page :wink:
Best regards
-
Bug: $file_ext resolves to file extension + " sign. :(
-
Bug: $file_ext resolves to file extension + " sign. :(
It doesn't do that here. What exactly are you doing, and what revision do you use?
-
I'm using exactly that "18 February 2008 build (4893)". I'm trying to integrate DJGPP's GCC. So I copied "GNU GCC Compiler" to "GNU GCC Compiler for DJGPP", noticed a too long command line for a DOS executable (gcc.exe) and changed $compiler ... -c $file ... to $compiler ... -c $file_name.$file_ext ... in advanced settings. Command line executed is gcc ... -c test.c" ....
Putting just $file_ext there also gives c".
-
Sorry, I still can't reproduce. My revision is 4887, but there have been no changes between these two revisions, other than include fixes and deleting old files. I couldn't imagine where that quote should come from either, as they are only added in pairs (and only when needed - test.cpp does not need).
This may be a stupid question, but I'm trying to rule out... your filename doesn't happen to contain a quote character? :)
To make sure I understood right what you are doing, this is what I did to reproduce:
(http://img176.imageshack.us/img176/6658/cmpwa9.png)
-
I already double checked the file name. It doesn't contain quotes. Did you ever try that? 8) Windows 2000 doesn't allow this.
I'm using exactly the same command line macro like you. Now I tried previous build 4872, but get the same problem.
Maybe it's related to my Windows 2000 Pro (SP4 + latest updates)?
-
I already double checked the file name. It doesn't contain quotes. Did you ever try that? 8) Windows 2000 doesn't allow this.
I'm using exactly the same command line macro like you. Now I tried previous build 4872, but get the same problem.
Maybe it's related to my Windows 2000 Pro (SP4 + latest updates)?
I just tested it on my W2K-Box with SP4 and latest updates.
It works fine, but I have to put $file_dir\ in front of $file_name to make it compile, because the sourcefile I used does not belong to any project.
-
Problem with rev4893 when using gdb and displaying a Disassembly of my code.
In rev4893 the disassembly window appears, but the eip position cursor is not shown, and the code starts at the beginning of the function, not at the set breakpoint / eip position.
I am using Windows XP SP2, gcc3.45, gdb 6.3. There are no spaces or quote in the filename/path.
I have reverted back to rev4875 and all is ok. I've not seen this problem in earlier versions.
RED
-
I just tested it on my W2K-Box with SP4 and latest updates.
It works fine, but I have to put $file_dir\ in front of $file_name to make it compile, because the sourcefile I used does not belong to any project.
I wiped all codeblocks settings from my HDD and tried again. Same problem. With $file_dir\ I get the following strange shot.
[attachment deleted by admin]
-
There seem to be problems with the whitespaces in the path.
To verify that, you can put your sources in a path without spaces and check it again.
-
People still keep adding bug reports (many of which refer to the exact same issues) even today. What is this telling you?
There is a filter needed.... :D
-
There seem to be problems with the whitespaces in the path.
To verify that, you can put your sources in a path without spaces and check it again.
Erm... very strange! Moving sources to a path w/o spaces works. No more ". Moving back to a path w/ spaces gives " again, but when I disable Use Full Paths for source files (GDB workaround) " disappears. Looks like a bug to me.
Btw: How does CB capture gcc's output for error parsing? (Doesn't work for me. :()
-
Afaik it parses it with regular expressions.
You find the used ones also under "advanced options".
It does not reliably work with localized compilers.
German gcc for example writes "Warnung:" instead of "Warning:", so every warning is treated as error.
I fixed the appropriate regexp on my system.
But I also tried to make printing, page-setup and print-previewing work correctly the last two weeks and needed a long time to figure out that some errors I had are because wxWidgets parses outputs (e.g. from lpstat on Linux) with hardcoded english strings (device etc).
So I think C::B's way with configurable regexp's is mutch more flexible.
-
Why is CB randomly adding bookmarks and ellipses to my gutter when i open a project?
-
Why is CB randomly adding bookmarks and ellipses to my gutter when i open a project?
BrowseTracker plugin does this.
-
Afaik it parses it with regular expressions.
I've already seen those regexps, but does CB redirect gcc's output to a file for parsing? When I take a failed command line from CB and run it directly from cmd.exe, I get a lot of error messages. Then I copy such a message to, e.g., "Compiler error" - "Test parsing expressions" and it is parsed correctly. But it is not displayed at "Build messages" tab. My guess is that CB doesn't "see" gcc's output.
-
Directive
#include "../blahblah.h"
appears not to work any more and results in a file does not exist error on compile. This only happens when path starts with "../
The files definitely exists, access rights are correct and they can be manually loaded but error persists during compile
Edit - the above now works with SVN 4908. Issue below still exists.
As a related issue one cannot remove files from project. The popup with the file names never comes up so I have to resort to editing .cbp file. This used to be a problem then got fixed but now is back.
Using SUSE 10.2 SVN 490X
-
Why is CB randomly adding bookmarks and ellipses to my gutter when i open a project?
BrowseTracker plugin does this.
ah, that one is placing those "...", but bookmarks ??
-
ah, that one is placing those "...", but bookmarks ??
It seems so. Disabling the plugin removed those.
-
ah, that one is placing those "...", but bookmarks ??
BrowseTracker is one of my favorite plugins, and I am very glad that it has been added to the contributed plugins list.
However, I have seen several questions about the "..." . Perhaps it would be preferable to have its Browse_Marks/Book_Marks turned off or hidden by default.
-
ah, that one is placing those "...", but bookmarks ??
BrowseTracker is one of my favorite plugins, and I am very glad that it has been added to the contributed plugins list.
However, I have seen several questions about the "..." . Perhaps it would be preferable to have its Browse_Marks/Book_Marks turned off or hidden by default.
Done. Svn 4916.
But it means current users of Browse Marks must now go enable BrowseMarks in the settings dialog.
-
BrowseTracker ... . Perhaps it would be preferable to have its Browse_Marks/Book_Marks turned off or hidden by default.
Done. Svn 4916.
But it means current users of Browse Marks must now go enable BrowseMarks in the settings dialog.
Pecan,
I updated and built SVN 4916 and the contributed plugins. Based on a couple of quick tests, it appears your BrowseTracker changes work as expected (on Windows XP).
-
Bug: $file_ext resolves to file extension + " sign. :(
I've seen this bug pop up with some batch scripts I wrote a while back. It happened when I took the scripts I had written, and copied them to another computer. Of course, I assumed they would work. They didn't.
As it turns out, the code that windows uses to generate a file name and pass it along to a program is fairly complex, and varies with the version of windows(which as you may not know, is not the same as the edition). In general:
1. if the file path contains spaces, the path is put inside quotes
2. if the file path doesn't contain spaces, the behavior is undefined(just like what a c++ compiler does for any number of ambiguous statements: i++ / 2 - ++i * 5)
From what I've seen, non-english windows tends to enclose filepaths in quotes more often than not. As for why you see it as a single quote at the end of a make variable, seems to me that the parsing for that variable needs a bit more error-checking. Took me 2 days to figure out what was going on with those batch scripts...
-
The version for MAC is not working with Spanish keyboards. I suppose also that it doesn't work over non-English keyboards. We can not have access to "[" or "{".
Also, I suppose the problem is in Scintilla.
Any idea?
-
The version for MAC is not working with Spanish keyboards. I suppose also that it doesn't work over non-English keyboards. We can not have access to "[" or "{".
Also, I suppose the problem is in Scintilla.
Any idea?
It works with my swedish keyboard, but you could be right...
-
The version for MAC is not working with Spanish keyboards. I suppose also that it doesn't work over non-English keyboards. We can not have access to "[" or "{".
Also, I suppose the problem is in Scintilla.
Any idea?
It works with my swedish keyboard, but you could be right...
I'm working with Code::Blocks since it s first beta version. It works on Windows perfectly, only it is working bad on Mac.
So I believe the problem will be on Scintilla/MAC
-
I meant it is working OK on Mac OS X 10.4 here
-
I meant it is working OK on Mac OS X 10.4 here
I supossed so, but the problem I have is only with two keys ( "[" and "{" ).
I suppose Swidish keyboard has more combinations that Spanish, but the rest of our keys is working perfectly. I'm must to investigate....
-
Hi,
I have a small problem... ;)
there is a bookmark in the editor window, marked as a blue triangle at the left side of the code...
when I remove this bookmark, close the file and reopen the file, the bookmark is there again !!! :shock:
is this a bug, or is there a specific way to remove a bookmark permanently ???
I'm using C::B from this thread on WinXP-pro (sp2)...
-
You might want to try the stable release 8.02 and then if the problem persists, post it in the appropriate forum (HELP) for example ~ good luck
-
Hi,
I have a small problem... ;)
there is a bookmark in the editor window, marked as a blue triangle at the left side of the code...
when I remove this bookmark, close the file and reopen the file, the bookmark is there again !!! :shock:
is this a bug, or is there a specific way to remove a bookmark permanently ???
I'm using C::B from this thread on WinXP-pro (sp2)...
I was unable to re-create this situation with the current SVN or stable CB 8.02.
Yes, when using the BrowseTracker plugin, book marks are reloaded when the file is simple closed and re-opened. This is so you do not lose your book marks when accidently closing the file, but the project is still open.
However, it should not be re-instating cleared Browse/Book/Marks.
I have seen this happen when the line was both BrowseMarked *and* BookMarked, *and* the user set BookMarks back to BrowseMarks via the settings menu. I was never able to catch it again.
The status of the Browse/Book/Marks are re-written when you close the project to <projectName>.bmarks .
BookMarks are reloaded by the BrowseTracker plugin when the user loads a project containing either BrowseMarks or Bookmarks and BrowseMarks are enabled. So if the BookMark is missing on project close, it's gone until the user re-toggles it.
In Stable CB 8.02, Book/Browse/Marks saving/loading has been disabled by default. Updating to that version, or updating to the current SVN should alleviate your problem.
-
ohh, sorry, I didn't see there's a new stable release !!! :oops:
it seems there is no problem with this version...
nm
-
Report bug: help plugin don't add right files in the db file when add more then 2 or 3 in same time. :( , i don't know why ?
AmR EiSa
-
Report bug: help plugin don't add right files in the db file when add more then 2 or 3 in same time. :( , i don't know why ?
AmR EiSa
Yup, you're right. I'll probably work on it tomorrow... if I don't forget.
-
Yup, you're right. I'll probably work on it tomorrow... if I don't forget.
Tssss... still.... :lol:
-
Devs,
I hope you don't mind my asking but It's planned to re-start the nightly builds (can be weekly or monthly..). Just to stay tuned with the latest SVN and make bug reports on the lastest SVN.
Max
-
Devs,
I hope you don't mind my asking but It's planned to re-start the nightly builds (can be weekly or monthly..). Just to stay tuned with the latest SVN and make bug reports on the lastest SVN.
Max
yes yes, was hoping of doing one today, but too jetlagged, will resume tomorrow :-)
-
Hello, I found a weird bug that is hard to reproduce. Well, maybe not after all.
When my mouse is leaving the editor window (the place where you type the code), the tool-tip will display what was under my cursor before I leave the editor window (not the Code::Blocks window, the text-area). As if it were detecting the mouse is still over the last position it was over. Anyways, this is a minor bug, but it still bugs me. =P
-- Wolf --
N.B.: I just noticed the links at the bottom to MySQL, PHP, W3C XHTML 1.0 and W3C CSS... "This page is not Valid XHTML 1.0 Transitional!" and the CSS generated errors. =P
-
Report bug: the language for input will be the 2nd. I have "eng & ara", C::B get ara for loc and use it not eng, and eng is the 1st :shock:.
AmR EiSa