# Help! Error Again Assertion \`no \< 100’ failed in igraph 0.8.3

**URL:** <https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487>\
**Category:** Usage\
**Tags:** C\
**Created:** [19 October 2020 16:44 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487 "2020-10-19T16:44:35Z")\
**Posts on this page:** 12\
**Page:** 1

<div class="post-metadata">

**Author:** ![d9y1kk](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/d9y1kk/32/382_2.png) [@d9y1kk](https://igraph.discourse.group/u/d9y1kk)\
**Post date:** [19 October 2020 16:44 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/1 "2020-10-19T16:44:35Z")

</div>

Hi, I asked a similar problem days ago and the error is solved by updating the igraph to 0.8.3.

Yesterday I change my code and this error appear again, I try to minimize the problem and reproduce it with a simple iterative graph building code and I attach it here for your testing

**Environment:** Visual Studio 2019 16.7.3, std=c++14, igraph 0.8.3

**Problem:** I failed at gCount=44, or the 45 rounds, exactly same error with this: [Need help with igraph error:[igraph\_error.c:186: Assertion `no \< 100’ failed.’]](https://igraph.discourse.group/t/need-help-with-igraph-error-igraph-error-c-assertion-no-100-failed/478). but with 0.8.3:

> Assertion failed: no \< 100, file D:\igraph-0.8.3-msvc\src\igraph\_error.c, line 186

**sample code:** run igraph\_debug\_test(). Sorry I extract the code from a long pipeline in a hurry, you may need OpenCV for it, or you can change my code to replace that cv::DMatch which is just a matching id storage object.

PS: If anyone can help adjusting the format, I try every option on editor none of them format the code properly.

```auto
bool extend1to2(igraph_t& sourceGraph, igraph_t& extendGraph, std::vector <cv::DMatch>& matches) {

	if (matches.size() == 0) {
		std::cerr << "graph.extend1to2: matches size is zero, invalid extention! return source graph instead!\n";
		return true;
	}

	int n_source_vertices = GAN(&sourceGraph, "n_vertices");
	int n_exd_vertices = GAN(&extendGraph, "n_vertices");
	std::unique_ptr<igraph_vector_t, void(*)(igraph_vector_t*)> 
		source_label(new igraph_vector_t(), &igraph_vector_destroy),
		exd_label(new igraph_vector_t(), &igraph_vector_destroy),
		source_edge_weight(new igraph_vector_t(), &igraph_vector_destroy),
		exd_edge_weight(new igraph_vector_t(), &igraph_vector_destroy);

	igraph_vector_init(source_label.get(),0);
	igraph_vector_init(exd_label.get(), 0);
	igraph_vector_init(source_edge_weight.get(), 0);
	igraph_vector_init(exd_edge_weight.get(), 0);
	VANV(&sourceGraph, "label",source_label.get());
	VANV(&extendGraph, "label", exd_label.get());
	EANV(&sourceGraph, "weight", source_edge_weight.get());
	EANV(&extendGraph, "weight", exd_edge_weight.get());

	igraph_t union_graph;
	igraph_disjoint_union(&union_graph, &sourceGraph, &extendGraph);
	igraph_vector_append(source_label.get(), exd_label.get());
	igraph_vector_append(source_edge_weight.get(), exd_edge_weight.get());
	SETVANV(&union_graph, "label", source_label.get());
	SETEANV(&union_graph, "weight", source_edge_weight.get());
	igraph_vector_t	contract_vertices_list;
	std::vector<igraph_real_t> vertices_ids(igraph_vcount(&union_graph),-1);
	for (int k = 0; k < matches.size(); k++) {
		vertices_ids[matches[k].queryIdx + n_source_vertices] = matches[k].trainIdx;
	}
	int count = 0;
	for (int i = 0; i < vertices_ids.size(); i++) {
		if (vertices_ids[i] == -1.0) {
			vertices_ids[i] = count;
			count++;
		}
	}

	igraph_vector_view(&contract_vertices_list, vertices_ids.data(), vertices_ids.size());
	igraph_attribute_combination_t comb;
	igraph_attribute_combination(&comb,
		"weight", IGRAPH_ATTRIBUTE_COMBINE_SUM,
		"", IGRAPH_ATTRIBUTE_COMBINE_FIRST,
		IGRAPH_NO_MORE_ATTRIBUTES);
	igraph_contract_vertices(&union_graph, &contract_vertices_list, &comb);
	igraph_simplify(&union_graph, 1, 1, &comb);
	SETGAN(&union_graph, "n_vertices", igraph_vcount(&union_graph));
	SETGAS(&union_graph, "name", GAS(&sourceGraph, "name"));
	sourceGraph = union_graph; 
}

bool buildFull(std::vector<DMatch>& matches, std::vector<KeyPoint>& kpts, igraph_t& mygraph) {

	igraph_init::attri_init();
	igraph_integer_t n_vertices = matches.size();
	igraph_bool_t loops = false;
	if (matches.size() == 0) {
		igraph_empty(&mygraph, 0, IGRAPH_UNDIRECTED);
		SETGAN(&mygraph, "n_vertices", 0);
		SETGAS(&mygraph, "name", "UNDEFINED");
		std::cout << "graph.build: warning: zero matches empty graph returned" << std::endl;
		return true;
	}

	int status = igraph_full(&mygraph, n_vertices, false, loops);
	SETGAS(&mygraph, "name", "fullgraph"); 
	SETGAN(&mygraph, "n_vertices", n_vertices); 

	if (status == IGRAPH_EINVAL) {
		std::cout << "build full graph: Invalid number of graph vertices" << std::endl;
		return false;
	}

	std::vector<igraph_real_t> labs(n_vertices);
	igraph_real_t* labels = labs.data();
	for (size_t i = 0; i < n_vertices; i++) {
		labs[i] = matches[i].trainIdx;
	}

	igraph_vector_t lab_vec;
	igraph_vector_view(&lab_vec, labels, n_vertices);
	
	SETVANV(&mygraph, "label", &lab_vec);

	igraph_simplify(&mygraph, 1, 1, 0);
	int n_edges = igraph_ecount(&mygraph);
	std::vector<igraph_real_t> w(n_edges);
	igraph_real_t* weights = w.data();
	for (size_t i = 0; i < n_edges; i++) {
		w[i] = 1.0;
	}
	igraph_vector_t weight_vec;
	igraph_vector_view(&weight_vec, weights, n_edges);
	SETEANV(&mygraph, "weight", &weight_vec);
	igraph_attribute_combination_destroy(&comb);*/
	return true;
}

void igraph_debug_test() {

	int gCount = 0;
	while (gCount != 50) {
		for(int i=0;i<1;i++){
			int total_num_feats_i = 100;
			std::vector<cv::DMatch> dbmatches;
			for (int i = 0; i < 100; i++) {
				dbmatches.push_back(cv::DMatch(i, i, -1, 0));
			}
			igraph_t base_graph;
			std::vector<cv::KeyPoint> pts;
			buildFull(dbmatches, pts, base_graph);
			if (GAN(&base_graph, "n_vertices") == 0) {
				std::cerr << "nbhdGraph.Next: error: base graph is empty!\n";
				break;
			}
			//extend the graph
			for(int j=0;j<2;j++){
				std::vector<cv::DMatch> exdmatches;
				for (int k = 0; k < 100; k++) {
					exdmatches.push_back(cv::DMatch(k, k + 80, -1, 0));
				}
				igraph_t exd_graph;
				std::vector<cv::KeyPoint> pts;
				buildFull(exdmatches, pts, exd_graph);
				if (GAN(&exd_graph, "n_vertices") == 0) {
					std::cerr << "nbhdGraph.Next: error: extend graph is empty!\n";
					break;
				}
				std::vector < cv::DMatch> newMatches;
				newMatches.reserve(20);
				for (int k = 0; k < 20; k++) {
					newMatches.push_back(cv::DMatch(k, k + 80, -1, 0));
				}
				if (!extend1to2(base_graph, exd_graph, newMatches)) {
					std::cerr << "nbhdGraph: error: graph extention failed! with img id " << i << "\n";
					break;
				}
				igraph_destroy(&exd_graph);
			}
			igraph_destroy(&base_graph);
		}
		std::cout << "\ngcount: " << gCount;
		gCount++;
	}
	return;
}

```

---

<div class="post-metadata">

**Author:** ![d9y1kk](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/d9y1kk/32/382_2.png) [@d9y1kk](https://igraph.discourse.group/u/d9y1kk)\
**Post date:** [21 October 2020 13:27 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/2 "2020-10-21T13:27:22Z")

</div>

Hello, is there any update? can you confirm that this is bug from igraph? If the bug fixing is time consuming, I may have to consider an alternative solution for my project.

---

<div class="post-metadata">

**Author:** ![vtraag](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/vtraag/32/38_2.png) [@vtraag](https://igraph.discourse.group/u/vtraag)\
**Post date:** [21 October 2020 14:04 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/3 "2020-10-21T14:04:10Z")

</div>

It is possible that this does constitute a bug. If you could report a [Minimal Working Example](https://stackoverflow.com/help/minimal-reproducible-example) that reproduces the problem, that would greatly help in identifying a possible bug. At the moment your code is a bit long, and I haven’t yet managed to get it to run after installing OpenCV.

Edit: it now seems I got OpenCV running (you are using OpenCV things without the namespace, and the poper includes are missing, I think `opencv2/core.hpp` is sufficient). However, the code you posted is missing `igraph_init::attri_init();`.

---

<div class="post-metadata">

**Author:** ![d9y1kk](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/d9y1kk/32/382_2.png) [@d9y1kk](https://igraph.discourse.group/u/d9y1kk)\
**Post date:** [21 October 2020 14:20 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/4 "2020-10-21T14:20:14Z")

</div>

sorry, it is a initalization:  
` igraph_i_set_attribute_table(&igraph_cattribute_table).`  
I will simplify my code now.

---

<div class="post-metadata">

**Author:** ![vtraag](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/vtraag/32/38_2.png) [@vtraag](https://igraph.discourse.group/u/vtraag)\
**Post date:** [21 October 2020 14:23 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/5 "2020-10-21T14:23:43Z")

</div>

> [@d9y1kk](#):
>
> sorry, it is a initalization:  
> ` igraph_i_set_attribute_table(&igraph_cattribute_table).`

Using that (and removing the line with `igraph_attribute_combination_destroy(&comb);*/`) I get the output:

```auto
nbhdGraph: error: graph extention failed! with img id 0

gcount: 0nbhdGraph: error: graph extention failed! with img id 0

gcount: 1nbhdGraph: error: graph extention failed! with img id 0

gcount: 2nbhdGraph: error: graph extention failed! with img id 0

gcount: 3nbhdGraph: error: graph extention failed! with img id 0

```

I doubt that this was the intention?

> [@d9y1kk](#):
>
> I will simplify my code now.

This will be very helpful to track down the (possible) bug.

---

<div class="post-metadata">

**Author:** ![d9y1kk](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/d9y1kk/32/382_2.png) [@d9y1kk](https://igraph.discourse.group/u/d9y1kk)\
**Post date:** [21 October 2020 14:50 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/6 "2020-10-21T14:50:15Z")

</div>

Sorry, I found that my first test code is definitely not readable. I created that code inside my project and that definitely a mistake.  
I write the new code in a new project and it should only depends on igraph now.

I simplify it as far as I can think:

```auto
#include <iostream>
#define IGRAPH_STATIC 1
#include "igraph.h"
#include <vector>
struct Dmatch{
	Dmatch(int query_Idx, int train_Idx) { queryIdx = query_Idx, trainIdx = train_Idx; }
	int trainIdx;
	int queryIdx;
};
bool extend1to2(igraph_t& sourceGraph, igraph_t& extendGraph, std::vector <Dmatch>& matches) {

	int n_source_vertices = GAN(&sourceGraph, "n_vertices");
	int n_exd_vertices = GAN(&extendGraph, "n_vertices");
	std::unique_ptr<igraph_vector_t, void(*)(igraph_vector_t*)>
		source_label(new igraph_vector_t(), &igraph_vector_destroy),
		exd_label(new igraph_vector_t(), &igraph_vector_destroy),
		source_edge_weight(new igraph_vector_t(), &igraph_vector_destroy),
		exd_edge_weight(new igraph_vector_t(), &igraph_vector_destroy);

	igraph_vector_init(source_label.get(), 0);
	igraph_vector_init(exd_label.get(), 0);
	igraph_vector_init(source_edge_weight.get(), 0);
	igraph_vector_init(exd_edge_weight.get(), 0);
	VANV(&sourceGraph, "label", source_label.get());
	VANV(&extendGraph, "label", exd_label.get());
	EANV(&sourceGraph, "weight", source_edge_weight.get());
	EANV(&extendGraph, "weight", exd_edge_weight.get());

	igraph_t union_graph;
	igraph_disjoint_union(&union_graph, &sourceGraph, &extendGraph);
	igraph_vector_append(source_label.get(), exd_label.get());
	igraph_vector_append(source_edge_weight.get(), exd_edge_weight.get());
	SETVANV(&union_graph, "label", source_label.get());
	SETEANV(&union_graph, "weight", source_edge_weight.get());
	igraph_vector_t	contract_vertices_list;
	std::vector<igraph_real_t> vertices_ids(igraph_vcount(&union_graph), -1);
	for (int k = 0; k < matches.size(); k++) {
		vertices_ids[matches[k].queryIdx + n_source_vertices] = matches[k].trainIdx;
	}
	int count = 0;
	for (int i = 0; i < vertices_ids.size(); i++) {
		if (vertices_ids[i] == -1.0) {
			vertices_ids[i] = count;
			count++;
		}
	}

	igraph_vector_view(&contract_vertices_list, vertices_ids.data(), vertices_ids.size());
	igraph_attribute_combination_t comb;
	igraph_attribute_combination(&comb,
		"weight", IGRAPH_ATTRIBUTE_COMBINE_SUM,
		"", IGRAPH_ATTRIBUTE_COMBINE_FIRST,
		IGRAPH_NO_MORE_ATTRIBUTES);
	igraph_contract_vertices(&union_graph, &contract_vertices_list, &comb);
	igraph_simplify(&union_graph, 1, 1, &comb);
	SETGAN(&union_graph, "n_vertices", igraph_vcount(&union_graph));
	SETGAS(&union_graph, "name", GAS(&sourceGraph, "name"));
	sourceGraph = union_graph;
}

bool buildFull(std::vector<Dmatch>& matches, igraph_t& mygraph) {
	igraph_i_set_attribute_table(&igraph_cattribute_table);
	igraph_integer_t n_vertices = matches.size();
	igraph_bool_t loops = false;

	int status = igraph_full(&mygraph, n_vertices, false, loops);
	SETGAS(&mygraph, "name", "fullgraph");
	SETGAN(&mygraph, "n_vertices", n_vertices);

	std::vector<igraph_real_t> labs(n_vertices);
	igraph_real_t* labels = labs.data();
	for (size_t i = 0; i < n_vertices; i++) {
		labs[i] = matches[i].trainIdx;
	}

	igraph_vector_t lab_vec;
	igraph_vector_view(&lab_vec, labels, n_vertices);

	SETVANV(&mygraph, "label", &lab_vec);

	igraph_simplify(&mygraph, 1, 1, 0);
	int n_edges = igraph_ecount(&mygraph);
	std::vector<igraph_real_t> w(n_edges);
	igraph_real_t* weights = w.data();
	for (size_t i = 0; i < n_edges; i++) {
		w[i] = 1.0;
	}
	igraph_vector_t weight_vec;
	igraph_vector_view(&weight_vec, weights, n_edges);
	SETEANV(&mygraph, "weight", &weight_vec);
	return true;
}

int main() {
	int gCount = 0;
	while (gCount != 50) {
		for (int i = 0; i < 1; i++) {
			int total_num_feats_i = 100;
			std::vector<Dmatch> dbmatches;
			for (int i = 0; i < 100; i++) {
				dbmatches.push_back(Dmatch(i, i));
			}
			igraph_t base_graph;
			buildFull(dbmatches,base_graph);
			if (GAN(&base_graph, "n_vertices") == 0) {
				break;
			}
			//extend the graph
			for (int j = 0; j < 2; j++) {
				std::vector<Dmatch> exdmatches;
				for (int k = 0; k < 100; k++) {
					exdmatches.push_back(Dmatch(k, k + 80));
				}
				igraph_t exd_graph;
				buildFull(exdmatches, exd_graph);
				std::vector <Dmatch> newMatches;
				newMatches.reserve(20);
				for (int k = 0; k < 20; k++) {
					newMatches.push_back(Dmatch(k, k + 80));
				}
				if (!extend1to2(base_graph, exd_graph, newMatches)) {
					break;
				}
				igraph_destroy(&exd_graph);
			}
			igraph_destroy(&base_graph);
		}
		std::cout << "\ngcount: " << gCount;
		gCount++;
	}
	return 0;
}

```

my result is also listed below:

 ![屏幕截图 2020-10-21 164933](https://global.discourse-cdn.com/free1/uploads/igraph/original/1X/ffed7ac467d96ec249ddcffa2d058fc74be21177.png)

Edit:  
I can explain what I’m doing here: I create a source fully-connected undirected graph with 100 vertices, and vertices with “label” 1-100, each edge with weight 1. Then I create a same type exd\_graph with 100 vertices, with “label” 80-179 and edge weight 1. Then the matching should be a 20 vector with vertices id [80-100].

Then I union the two graphs, contract the vertices with corresponding ids, and add edge weight together, which is defined in the igraph combination object. I repeat twice and extend two graphs to the source one. At final, I delete all graphs.

I repeat this process 50 times, it crashed at 44 round.

---

<div class="post-metadata">

**Author:** ![vtraag](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/vtraag/32/38_2.png) [@vtraag](https://igraph.discourse.group/u/vtraag)\
**Post date:** [21 October 2020 18:05 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/7 "2020-10-21T18:05:47Z")

</div>

Thank you for reporting this, it indeed revealed a bug.

I opened an issue on GitHub [here](https://github.com/igraph/igraph/issues/1484) and pushed a fix in [PR #1485](https://github.com/igraph/igraph/pull/1485). If you really need the corrected code immediately, and you cannot wait before it is merged, you can simply fetch the code from my [fix branch](https://github.com/vtraag/igraph/tree/fix/1484).

---

<div class="post-metadata">

**Author:** ![d9y1kk](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/d9y1kk/32/382_2.png) [@d9y1kk](https://igraph.discourse.group/u/d9y1kk)\
**Post date:** [21 October 2020 18:46 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/8 "2020-10-21T18:46:21Z")

</div>

@vtraag  
Edit: Test with your branck ( `igraph-fix-1484`)  
same error with attribute query function VAN and report attributes lost.

Hi, thanks for the fix. I think this is for the latest branch right? I check the fix and it is three line changes in cattributes.c:

> //IGRAPH\_FINALLY\_CLEAN(2);  
> igraph\_i\_cattribute\_permute\_free(new\_val);  
> IGRAPH\_FINALLY\_CLEAN(3);

so I change the source code and recomplie, but another error is occurred:

> Error at D:\thesis\igraph-0.8.3-msvc\src\cattributes.c:2381 :Unknown attribute, Invalid value

error is at the line:

> [@d9y1kk](#):
>
> `VANV(&sourceGraph, "label", source_label.get());`

I track the bug, the code failed at the second graph extention (j=1), which I guess `igraph_contract_vertices` in the j=0 loop may fail to keep the vertices “label” attributes after contraction.

of course I’m in the release version 0.8.3, I will test you version.

---

<div class="post-metadata">

**Author:** ![vtraag](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/vtraag/32/38_2.png) [@vtraag](https://igraph.discourse.group/u/vtraag)\
**Post date:** [21 October 2020 20:37 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/9 "2020-10-21T20:37:05Z")

</div>

This is rather unexpected. I tested the exact code you provided, and this works without any problems with the fix. I have tested up to `gCount == 500`, and didn’t encounter any problem. Perhaps you want to double check whether everything is correctly compiled.

By the way, the PR is merged in the `master` branch of the `igraph` repo, so you can also simply pull from there. Just to make sure: when building from GitHub the instructions are slightly different: see [here](https://igraph.org/c/compilation-windows.html) for instructions.

It might be that this is a specific issue for Windows (I am working on Ubuntu myself), but I doubt whether that is the case.

---

<div class="post-metadata">

**Author:** ![d9y1kk](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/d9y1kk/32/382_2.png) [@d9y1kk](https://igraph.discourse.group/u/d9y1kk)\
**Post date:** [21 October 2020 20:46 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/10 "2020-10-21T20:46:25Z")

</div>

Yes, I followed the instruction and build the msvc project from the code, all works.

Besides, I have tried reverse the code ( comment the new lines and use the IGRAPH\_FINALLY\_CLEAN(2)) , the same IGRAPH\_FINALLY\_CALL error can be reproduced and crashed at 44th round, which means the VAN works normally and these two lines may accidently release the attributes? This might still be a igraph problem.

---

<div class="post-metadata">

**Author:** ![vtraag](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/vtraag/32/38_2.png) [@vtraag](https://igraph.discourse.group/u/vtraag)\
**Post date:** [21 October 2020 21:06 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/11 "2020-10-21T21:06:19Z")

</div>

Hmm, you might actually be right. It seems that the fix actually removed the new values. I am not sure why I didn’t receive an error on my end, but this might just be luck.

Could you please check my [fix branch](https://github.com/vtraag/igraph/tree/fix/1484) again?

---

<div class="post-metadata">

**Author:** ![d9y1kk](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/d9y1kk/32/382_2.png) [@d9y1kk](https://igraph.discourse.group/u/d9y1kk)\
**Post date:** [21 October 2020 21:19 UTC](https://igraph.discourse.group/t/help-error-again-assertion-no-100-failed-in-igraph-0-8-3/487/12 "2020-10-21T21:19:47Z")

</div>

Change the source code according to your fix, test on the example code and on my project, works perfect, thanks. Great to know it is finally fixed, I don’t need to rewrite my whole project now 😄