Can't find compiler - Page 1

I've got this error today. Last week, I have installed TIGCC0.96b8, because Godzil's site was down for a few moment. I hadn't see vince's mirror.
Today, I've installed GCC4TI instead of TIGCC. Process :
- removed TIGCC
- installed GCC4TI
- used CCleaner to remove TIGCC entries in the register

=> I got an error (compiler not found) when I tried to compile.

- reinstalled GCC4TI

=> same error

- reinstalled TIGCC
- recompiled with GCC4TI

=> success

Did I something wrong ? Perhaps CCleaner has broken something ?

The GCC4TI IDE uses the same registry keys as the TIGCC IDE (which is why I told you, in the topic about syntax coloring fixes, not to launch any other version of the IDE).

Thank you. But why didn't it work after having re-installed GCC4TI ?

Good question... I didn't modify the code in the IDE that reads or writes the registry keys ?!
Since the GCC4TI IDE and the TIGCC IDE are compatible, they use the same registry keys.

It couldn't be a problem with the installer ? Did you try to install GCC4TI on a PC without any previous TIGCC installation ?

No, I didn't try to install GCC4TI on a PC without any previous TIGCC installation...

But I didn't modify the code that reads or writes the registry keys in the IDE, or (I've just checked) in the installer:

~/gcc4ti/trunk/tigcc$ grep -Rin "SeReSoft" .
./ide/MainUnit.pas:531: RegKey = '\Software\SeReSoft\TI-GCC IDE';
./setup/tigcc.nsi:115:  ReadRegStr $0 HKCU "Software\SeReSoft\TI-GCC IDE" "Program Folder"
./setup/tigcc.nsi:116:  StrCmp $0 "" noseresoft
./setup/tigcc.nsi:119:  ReadRegStr $0 HKCU "Software\SeReSoft\TI-GCC IDE" "Program Group"
./setup/tigcc.nsi:120:  StrCmp $0 "" noseresoft
./setup/tigcc.nsi:122:noseresoft:
./setup/tigcc-cross.nsi:114:  ReadRegStr $0 HKCU "Software\SeReSoft\TI-GCC IDE" "Program Folder"
./setup/tigcc-cross.nsi:115:  StrCmp $0 "" noseresoft
./setup/tigcc-cross.nsi:118:  ReadRegStr $0 HKCU "Software\SeReSoft\TI-GCC IDE" "Program Group"
./setup/tigcc-cross.nsi:119:  StrCmp $0 "" noseresoft
./setup/tigcc-cross.nsi:121:noseresoft:
Fichier binaire ./doc/Programs/Edit/EditHelpFile.exe concorde

~/tigcc-cvs-co/tigcc$ grep -Rin "SeReSoft" .
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/ide/MainUnit.pas:528:   RegKey = '\Software\SeReSoft\TI-GCC IDE';
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc.nsi:113:  ReadRegStr $0 HKCU "Software\SeReSoft\TI-GCC IDE" "Program Folder"
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc.nsi:114:  StrCmp $0 "" noseresoft
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc.nsi:117:  ReadRegStr $0 HKCU "Software\SeReSoft\TI-GCC IDE" "Program Group"
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc.nsi:118:  StrCmp $0 "" noseresoft
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc.nsi:120:noseresoft:
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc-cross.nsi:112:  ReadRegStr $0 HKCU "Software\SeReSoft\TI-GCC IDE" "Program Folder"
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc-cross.nsi:113:  StrCmp $0 "" noseresoft
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc-cross.nsi:116:  ReadRegStr $0 HKCU "Software\SeReSoft\TI-GCC IDE" "Program Group"
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc-cross.nsi:117:  StrCmp $0 "" noseresoft
/home/Lionel_Debroux/tigcc-cvs-co/tigcc/setup/tigcc-cross.nsi:119:noseresoft:
Fichier binaire ./doc/Programs/Edit/EditHelpFile.exe concorde

This explains why the syntax coloring doesn't seem to be updated when switching from TIGCC IDE to the bugfixed GCC4TI IDE I talked about at topics/120340-syntax-highlighting .
But that doesn't explain why it didn't work for you after the clean reinstall of GCC4TI...

I'll retry :
-> remove TIGCC & GCC4TI
-> CCleaner
-> install TIGCC and test (already seems to work)
-> remove it + CCleaner
-> install GCC4TI and test

Thanks

Please also test the syntax coloring by doing:
-> remove TIGCC & GCC4TI
-> CCleaner
-> install TIGCC and check whether tas is colored (it shouldn't be)
-> remove it without using CCleaner
-> install GCC4TI and check whether tas and several others are colored (they're not expected to be, since the IDE stored coloring information that doesn't feature them)

Thanks in advance.

Ok.

You need to reset the syntax highlighting settings after upgrading the IDE to a version with fixed definitions, it has always been like that in TIGCC. That's what the Reset button in the syntax highlighting preferences is for. The reason it's done that way is that you may have customized the settings and your customizations shouldn't get overwritten.

I didn't know of the Reset button, indeed...

You need to reset the syntax highlighting settings after upgrading the IDE to a version with fixed definitions, it has always been like that in TIGCC. [...] The reason it's done that way is that you may have customized the settings and your customizations shouldn't get overwritten.

Of course customizations shouldn't be automatically overwritten by an IDE upgrade
At the very least, before attempting to do so, there should be a confirmation dialog.

But here, what we have is:
* upon upgrading the IDE to a version containing fixed definitions, no indication is given to the user that the syntax highlighting definitions are outdated (and I wonder where the indication about _how_ to upgrade is given, since I didn't know about it, and I'm not exactly a newcomer, though I have hardly used the IDE for a while);
* the Reset button kills the customizations without any form of warning or confirmation.
This is not very user-friendly.

The IDE configuration, which is a superset of the syntax highlighting/coloring definitions, is not versioned in the registry (thankfully, Tom didn't make that mistake when making the TICT programs that load & save things). That makes it harder to automatically merge user customizations into updated base definitions, since it's harder to detect which is the version of the base definitions...

Lionel Debroux (./11) :
* the Reset button kills the customizations without any form of warning or confirmation.

The Reset button only takes effect if you click OK in the dialog, that's confirmation.

BTW, I did last week an install of GCC4TI 0.96b10 on a fresh install of Win XP sp3, and all works fine.

I completely forgot about this topic (none of us created a ticket), but I stumbled across the bug myself, confirmed that it was a real bug, and fixed it as part of r1348.

Thanks for the follow-up nonetheless