I've set tag_count = 10 in my config, but I get the works space 2 and 4 at the same time when I go to workspace 10 with bind=SUPER,0,view,10,0 Super+0

I'm so impatient. 😅

Binary files 1/.git/index and 2/.git/index differ
diff -u -ruw 1/src/config/parse_config.h 2/src/config/parse_config.h
--- 1/src/config/parse_config.h	2026-03-06 17:38:04.940543509 +0100
+++ 2/src/config/parse_config.h	2026-03-06 17:38:17.577672294 +0100
@@ -1578,6 +1578,7 @@
 		config->default_nmaster = atoi(value);
 	} else if (strcmp(key, "tag_count") == 0) {
 		config->tag_count = CLAMP_INT(atoi(value), 1, 32);
+		tag_count = config->tag_count;
 	} else if (strcmp(key, "center_master_overspread") == 0) {
 		config->center_master_overspread = atoi(value);
 	} else if (strcmp(key, "center_when_single_stack") == 0) {
Only in 1/src/config: parse_config.h.orig
Only in 1/src/dispatch: bind_define.h.orig
diff -u -ruw 1/src/ext-protocol/ext-workspace.h 2/src/ext-protocol/ext-workspace.h
--- 1/src/ext-protocol/ext-workspace.h	2026-03-06 17:38:04.941927965 +0100
+++ 2/src/ext-protocol/ext-workspace.h	2026-03-06 17:38:17.580583016 +0100
@@ -106,7 +106,7 @@
 }
 
 static void add_workspace_by_tag(int32_t tag, Monitor *m) {
-	char name_buf[4];
+	char name_buf[12];
 	const char *name = get_name_from_tag(tag, name_buf, sizeof(name_buf));
 
 	struct workspace *workspace = ecalloc(1, sizeof(*workspace));

Hey @hboetes , did you restart the compositor after setting the tag_count config var? Also, are you sure you didn't already have another bind in your config for super,0 to enter overview mode? I'm not able to reproduce your issue. That said, your patch did lead to some changes; thanks for that. See my latest commit.

This has been stable for me, so I'm going to mark this PR as ready for review.

Thanks for incorporating the fix in the first hunk of my patch! One small note on name_buf: the CLAMP_INT in the snprintf call doesn't actually silence the -Wformat-truncation warning, because the compiler sees the theoretical range of a uint32_t (up to 10 digits), not the runtime-clamped value. The buffer just needs to be 12 bytes (10 digits + NUL + 1 spare) to make both the compiler and the logic correct. The CLAMP_INT can then be dropped as it's redundant, given the earlier tag > tag_count guard.

I didn't get that warning from either GCC or Clang, but I was generally unhappy with the get_name_from_tag function, so I took a pass at disambiguating the word tag (clarified whether it's a number in [1, tag_count]) and hardened the get_name_from_tag (now renamed to get_name_from_tag_number for clarity) function. The new signature should serve the needs of named workspaces in the future, at least while workspaces are still mapped to tags.

I took another pass at making the primitives for representing a tag more consistent to avoid future UB.

I also began simplifying the CLAMP macros. @DreamMaoMao - let me know what you think of these and I can either do the full refactor in this PR or do a separate one.

At the first restart I changed from 10 to 5, correct reply.

724:00:02:55.664 [INFO] [../src/config/parse_config.h:1605] tag_count change requires restart (current: 10, requested: 5)

And at this restart I changed the number of tags to -1, and nothing spectacular happened. Also good.

743:00:04:05.811 [INFO] [../src/config/parse_config.h:1605] tag_count change requires restart (current: 10, requested: 1)

PLEASE ADD THIS FEATURE

As I mentioned in another pr, instead of expanding the number of tags, it is better to use a more flexible dynamic tagging system. However, this workload is extremely huge, and I still have other more important work to do

I completely understand the desire for a proper dynamic tagging system, that would be the right long-term solution. Since that work is still on the horizon though, would it make sense to merge this as an interim measure? It can always be refactored once the dynamic system lands, and in the meantime it unblocks current users.

Thanks for your consideration.

Hi @DreamMaoMao , I've updated this PR to be compatible with the latest config.* refactor that got rid of the global variables. This PR is small, doesn't re-architect anything, doesn't break anything, doesn't get in the way of a future refactor, and people can immediately benefit from it. Please reconsider your position.

If you're absolutely against this PR and will only ever accept a change with a more flexible dynamic tagging system, then it would be helpful to share what you had in mind in the form of a design document or discussion. Then we can see who'd be willing to work on it, how long it'll take, testing effort, etc. Until then, this PR exists right now and I'm happy to make any changes you'd like.